Skip to main content

Full text of "Determination of horizontal stress in soils"

See other formats


TE 

662 

.A3 

no . 

FHWA- 

RD- 

Q l-I 18 



eport No. FHWA/RD-81/118 

lETERMINATION OF HORIZONTAL 
oTRESS IN SOILS 



August 1981 
Final Report 




DEP/ 
TRAr 



"MENT OF 

MUTATION 

.1 Ibdi 



LIBRARY 



rtSLSStoi 




^TES O* 



Document is available to the public through 
the National Technical Information Service, 
Springfield, Virginia 22161 



Prepared for 

FEDERAL HIGHWAY ADMINISTRATION 
Offices of Research & Development 
Materials Division 

Washington, D.C. 20590 



FOREWORD 



This report presents the results of a 2 -year study which led to 
the development of a lateral stress sensor and test methods for 
determining the horizontal stress in soils in-situ. 

The study included laboratory and field tests in several different 
soil types and comparisons between lateral stress measurements 
obtained with a self boring pressuremeter and the lateral stress 
sensor. The results indicate that reliable measurements of lateral 
stress can be made quickly and simply. 

The work reported resulted from FCP Project 5B Study, 35B2-552, 
"Measuring and Testing Techniques for Determination of the In-Situ 
State of Stress in Soils," conducted by Soils Systems, Inc., 
Marietta, Georgia. The research was performed under DOT-FH-11-9172 , 
during the period September 28, 1976, to March 31, 1979. 

Copies of the final report are being distributed by the Office of 
Research, Materials Division, to other researchers and to appropriate 
members of the FCP Project 5B Team. 




!/£*■ 



"es F. Scheffey 
Director, Office of Research 
Federal Highway Administration 



NOTICE 

This document is disseminated under the snonsorshio of the Department 
of Transportation in the interest of information exchange. The United 
States Government assumes no liability for its contents or use thereof. 
The contents of this report reflect the views of the contractor, who 
is responsible for the accuracy of the data presented herein. The 
contents do not necessarily reflect the official views or policy of 
the Department of Transportation. This reoort does not constitute a 
standard, specification, or regulation. 

The United States Government does not endorse products or manufacturers 
Trade or manufacturers' names anpear herein only because they are 
considered essential to the object of this document. 



; 






Technical Report Documentation Page 



1 . Report No. 

FHWA/RD-81/118 



2. Government Accession No. 



4. Title and Subtitle 

Determination of Horizontal Stress in Soils 



7. Author's) Dri N> s> Fox, Dr. R. L. Handy, Gary D. Trott 
Bernard Remmes, Steven Moldt 



9. Performing Organization Name and Address 

Soil Systems, Inc. 

525 Webb Industrial Drive 

Marietta, Georgia 30062 



12. Sponsoring Agency Name and Addres 



Offices of Research and Development 
Federal Highway Administration 
U. S. Department of Transportation 
Washington, D.C. 20590 



3. Recipient's Catalog No. 



5. Report Date 

August 1981 



6. Performing Organization Code 



i. Performing Organization Report No. 



10. Work Unit No. (TRAIS) 

35B2-552 



11. Contract or Grant No. 

DOT-FH-ll-9172 



13. Type of Report and Period Covered 

Final Report 



14. Sponsoring Agency Code 

M/0706 



Carl Ealy (HRS-21) 



15. Supplementary Notes 

FHWA Contract Manager: 

16. Abstract 

This research involves calculating the state of stress in a soil mass by 
recognizing that disturbance during testing or measuring is inevitable. 
Thin blades with teflon-diaphragm pneumatic s cress cells have been developed 
for this research to measure soil stresses. Soil stresses on the blade 
were found to be a function of blade thickness. An exponential relation- 
ship between blade thickness and soil stresses was determined to exist except 
in hard soils with thick (1/4 inch) blades. 



A three-bladed stepped vane with nine pressure cells ha 
designed, built, and tested in preliminary trials. Use 
will result in the measurement of a two dimensional soi 



Fressuremeter data from two test sites, one in Glacial till and loess, the 
other in expansive clay, indicated very close agreement with blade stress 
results. The blade was easier to use than the pressurer letSEP&RJMfe^Tau^e or 
its statistical advantage (many more test results per dc y )TPAif^I?iQBSAJKe'ci:ie. 



bd^E CsuoieB&liir; 

of this instrumen- 

stre lfBF?M e - 



17. Key Words 



18. Distribution Statement 

No restrictions. This document is 
available to the public through the 
National Technical Information Service, 
Springfield, Virginia 22161. 



19. Security Classif. (of this report) 



Unclassified 



20. Security Classif. (of this page) 

Unc] assif ied 



21- No. of Pages 22. Price 



205 



Form DOT F 1700.7 (8-72) 



Reproduction of completed page authorized 



TABLE OF CONTENTS 

Page 

INTRODUCTION 1 

SCOPE 9 

LITERATURE REVIEW 10 

Determination of Stresses in a Soil Mass 10 

Theoretical Determinations of Kq 11 

Laboratory Determinations 12 

In Situ Evaluation of Kq 15 

Penetration Mechanics 20 

Disturbance 20 

Wedge Shape 26 

Pore Pressure 27 

Stress Cell Location 28 

INVESTIGATIVE PROCEDURE 31 

Material Properties 31 

Sand 31 

Modeling Clay 37 

Blade Insertion into Sand 40 

Data Reduction 44 

Blade Angle 46 

Mold Effects 46 

Blade Insertion into Modeling Clay 50 

Data Reduction 52 

Rate of Penetration 53 

Blade Angle 54 

STRESS SENSOR DESIGN 55 

Model Test 55 

Procedure 55 

Preliminary Diaphragm Selection 63 



li 



TABLE OF CONTENTS (CONTINUED) 

Page 

Blade-Mounted Sensor 66 

Further Sensor Designs 69 

Calibration 70 

Peripheral Equipment 7 3 

LABORATORY TESTING 77 

Disturbance in Layered Specimens 77 

Stress Concentration 81 

Stress vs. Blade Thickness 85 

Exponential Curve Fit 99 

Stress Sensor Tests in K-Test Mold 103 

Stress Sensor Tests in Test Box 108 

FIELD IN SITU TESTS 114 

Methods and interpretation 114 

Single Blade Tests, Overconsolidated Loess 115 

Single Blade Tests, Underconsolidated Loess 120 

Single Blade Tests, Alluvium 123 

Stepped Blade Tests in FHWA Test Pit 125 

Stepped Blade Tests, Mitchellville Till Site 128 

Stepped Blade Tests, Houston Clay 133 
Stepped Blade Tests at Fairbank Highway Research 

Station, McLean, Virginia. 141 

CONCLUSIONS 152 

FIELD TESTING PROCEDURES 155 

REFERENCES CITED 156 

APPENDIX A 160 

APPENDIX B 177 

APPENDIX C 183 

APPENDIX D 197 



ill 



LIST OF FIGURES 

Page 

1. Methods for evaluating horizontal in situ soil stress. 2 

2. Stress relaxation times for various techniques of in situ 
horizontal stress measurement (after Tavenas (31) ) . 5 

3. Hypothetical example of extrapolation to initial in situ 

stress 6 

4. Conceptual drawing of a vane stress sensor 8 

5. Kq as a function of overconsolidation ratio (from Brooker 

and Ireland (10)) . 14 

6. "Camkometer" self-boring pressuremeter (from Wroth (42)). 18 

7. Foundation failure modes (from Vesic (37)). 22 

8. Assumed plastic zones under deep foundations (from Vesic 

(39)). 23 

9. Sand displacement around a pile (from Robinsky and 

Morrison (26) ) . 25 

10. Representation of pore pressure effects in soil. 29 

11. Grain size curve of sand used in this project. 32 

12. Shear strength test results for sand. 33 

13. The Iowa K-Test mold (from Lutenegger (18)). 35 

14. X-ray diffracto traces of modeling clay. 39 

15. Laboratory testing apparatus. 43 

16. Photographs of special strip blade. 48 

17. Test results from special strip blade in sand. 49 

18. Photograph of a layered modeling clay specimen with a metal 

blade inserted into it. 51 

19. Schematic of laboratory set-up used in testing gas-flow 
pressure sensor concept. 56 

20. Drawing of sensor simulator used to evaluate the pneumatic 

stress cell. 58 



IV 



LIST OF FIGURES (CONTINUED) 



Page 

21. Typical plot of results obtained from a steel diaphragm 61 
performance test. 

22. Photographs of Teflon diaphragm stress sensor. 71 

23. Calibration plots for the Teflon diaphragm stress sensors 

of the stepped VSS device. 72 

24. Photographs of blade with Teflon diaphragm stress sensors. 74 

25. Schematic of stress sensor pressurizing system. 75 

26. Measured disturbance in 40% relative density sand specimens. 79 

27. Measured disturbance in the low consistency clay specimen. 82 

28. Plots of disturbance as a function of location in the low 
consistency clay. 83 

29. Plots of disturbance as a function of location in the low 
consistency clay. 84 

30. Test results from 3 mm thick smooth blade inserted into a 

75% D r sand specimen. 86 

31. Results from a 1/8" thick smooth balde inserted into a low 
consistency clay specimen. 87 

32. Test results from 2 in. wide smooth blades inserted into 

40% D r sand specimen. 88 

33. Test results from 1 in. wide rough blades inserted into 

40% D r sand specimens. 89 



34. Test results from 1 in. wide smooth blades inserted into 
40% D r sand specimens . 



90 



35. Test results from 2 in. wide smooth blades inserted into 

75% D r sand specimens. 91 

36. Test results from 2 in. wide rough blades inserted into 75% 

D r sand specimens. 92 

37. Test results from 2" wide smooth blades inserted into low 
consistency clay specimens. 95 

38. Test results from 2" wide rough blades inserted into low 
consistency clay specimens. 96 



LIST OF FIGURES (CONTINUED) 

Page 

39. Test results from 2.5 cm smooth blades inserted into low 
consistency clay specimens. 97 

40. Test results from 5 cm wide smooth blades inserted into 

high consistency clay specimens. 98 

41. Sketch of test box. 110 

42. Test Site near Boone, Iowa. 117 

43. Soil data from the Boone, Iowa Test Site. 118 

44. Test site - Turin, Iowa. 121 

45. Soil Data, Logan-1 Test Site. 124 

46. Data, Logan-2 Test Site. . 126 

47. Soil Data, FHWA test pit. 129 

48. Soil Data, Mitchellville Test Site. 132 

49. Houston Clay stiffness factor "b" versus depth. 135 

50. Horizontal stress vs. depth in Houston clay. 138 

51. K versus depth for Houston clay. 142 

52. Fairbank Highway Research Station, McLean, Virginia site 

"b" values from grouped data. 145 

53. Horizontal stress vs. depth, Fairbank site. 146 

54. K Q versus depth for Fairbank site. 151 



VI 



LIST OF TABLES 

Page 

1. Summary of diaphragm test results 64 

2. Average properties of selected engineering materials 65 

3. Summary of test results using a blade with steel diaphragm 

stress sensor. 68 

4. List of sand tests used in extrapolation procedure. 93 

5. List of clay tests used in extrapolation procedure. 94 

6. Summary of best-fit exponential curves to stress ratio data. 100 

7. Summary of best-fit exponential curves to blade load data. 101 

8. Physical properties of laboratory test soils. 104 

9. Summary of single-blade laboratory tests. 106 

10. Results of stress sensor tests in test box. Ill 

11. Results of blade stress sensor tests at Boone site. 119 

12. Results of blade stress sensor tests, Turin site. 122 

13. Results of blade stress sensor tests at Logan-Test sites. 127 

14. Results of blade stress sensor tests at FHWA test pit and 
Mitchellville test sites. 130 

15. Soil data, University of Houston test site. 133 

16. Stress data, Houston clay. 136 

17. Houston clay stresses with linearly regressed b data. 139 

18. Houston clay Kq data. 140 

19. Soil data, FHWA Fairbank Highway Research Station, McLean, 
Virginia test site. 140 

20. Results of stepped blade tests, FHWA Fairbank Highway Research 
Station, McLean, Virginia test site. 143 

21. Grouped data calculations of b values for Fairbank site. 147 

22. Fairbank soil stresses with linearly regressed : 'b" data 148 

23. Fairbank Soil Kq data. 150 



VII 



INTRODUCTION 

Because of increasing emphasis on urban mass transit, in 1976 the 
Transportation Research Board prepared a special report (30) on tunnel 
construction. The report was a state-of-the-art paper, and recommended 
present and future research needs. One such need is to better predict 
loads on structural supports in tunnels. Up until now, design of 
supports has depended to a great extent on the ability of the engineer 
to convert limited geological data into practical information. If the 
engineer overestimates loads, the design is uneconomical; if he under- 
estimates, yielding in the structure can result. This yielding can 
cause excessive ground movement and increase underpinning requirements 
for adjacent structures. 

Being able to calculate the state of the stresses in a soil mass 
is of considerable importance to geotechnical engineers. For example, 
a safe and economical design of a pile or deep foundation requires that 
the engineer have a working knowledge of the in situ stresses in a 
soil mass. The accuracy with which an engineer estimates the lateral 
stresses that act upon a retaining wall will affect his final design. 
Use of the finite element method of analysis in geomechanics is 
severely limited by the engineer's ability to estimate initial in situ 
stresses. 

A number of schemes have been devised to measure in situ stresses 
in soils, illustrated in Figure 1. Hydraulic fracturing consists of 



Hydraulic 



Fracture 



en: 



\ 



y 






Self- boring 



Pres- 



suremeter 




"it 



^ 



°>». 




Glotzl 



Cell 



Flat Dilatometer 






0— r K^h 



\ 



t,-3 



tt 



fc* 



X^Jh^l Stepped Vane 

K 



%. 



-P(t) 



Figure 1. Methods for evaluating horizontal in situ soil stress. 



pumping water into a piezometer with gradually increasing pressure 
while monitoring the pumping rate, a sudden increase in flow being 
indicative of fracture in the direction of least resistence, i.e. 
tensile fracture in the direction of the minor principal stress. 
Errors arise from the disturbance effects from withdrawing a piezometer 
and the possibility for cavity expansion without fracture. Tensile 
strength of the soil is taken into account by cycling the pressure 
and monitoring "closure." 

The other devices for measuring in situ soil stress are essentially 
mechanical with either hydraulic or electrical transducers and readout. 
These follow one of three patterns: (a) Introduce a thin pressure 
cell so as to minimize disturbance, but recognize that there has been 
disturbance and repeat the pressure determinations over a long period of 
time so as to predict the final pressure at equilibrium: Glbtzel cell, 
(b) Introduce a thin pressure cell and expand it to measure soil 
response, then in effect by use of empirical data back-calculate for 
zero thickness: Flat Dilatometer. (c) Introduce a cylindrical pressure 
cell by making it integral with a soil drill so as to minimize soil 
disturbance and relaxation, then monitor pressure over a period of 
time: Self-Boring Pressuremeter . 

The problem with most of these methods is the waiting time to 

31 
reach equilibrium, shown by Tavenas to vary from a few hours for 

the self-boring pressuremeter to several years for hydraulic fracturing, 



Figure 2. Furthermore there is no assurance that the re-attained 

equilibrium represents the original stress state. 

The principle embodied in the research described herein is to 

recognize that disturbance is inevitable, then vary it in discreet 

steps, and determine the pressure as a function of disturbance so as 

to allow an extrapolation of pressure to zero disturbance. This is 

done with a stepped blade or vane, Figure 1: Pressures (J, , o\ , a 

h l h 2 h 3 
are measured after insertion of blades of thicknesses t.. , t„, and t„. 

If a is found to be some continuous function of blade thickness, 

extrapolation should be possible to find 0" , the pressure on a 

o 
hypothetical blade of zero thickness. The function need not be 

linear; it must only be continuous. An hypothetical example is shown 

in Figure 3 . 

We should note that extrapolation to zero thickness still is not 

the same as the true undisturbed stress state, since the infinitely 

thin blade still represents a discontinuity that does not pass through 

individual soil grains, but pushes them aside. This means a higher- 

than-original pressure may be determined, the error being largest 

for dense, granular soils that dilate upon blade intrusion. Dilation 

or volume increase during shear is a direct evidence for disturbance. 

We therefore may predict problems with dense sands, as indeed will be 
shown, but such error should be less in loose sands, and a minimum in 
soft or moderately firm silts and clays. 

It also may be seen that instead of pushing a single blade as in 
Figure 1, we may assemble several blades into a single instrument so as 



2.2 



co 
co 

w 
« 

Eh 
CO 

W 

< 

Eh 
CO 

Z 

M 
En 

•I* 
■P 



s 



Eh 

<: 

CO 
CO 

w 
« 

Eh 
CO 



2.0 - 



1.6 K 



1.4 _ 



1.2 _ 



1.0 



o Self-boring pressuremete]): 

A Camkometer 

• Total pressure cells 

V Hydraulic fracturing 




10 10 10 

TIME AFTER DRIVING, HOURS 



10 



10 



Fig. 2. Stress relaxation times for various techniques of in 
situ horizontal stress measurement (after Tavenas 31) 




FUNCTION OF BLADE THICKNESS 



Figure 3. Hypothetical example of extrapolation to 
initial in situ stress. 



to determine a two-dimensional stress state, Figure 4. This was the 
eventual goal of the research. 




Figure 4. Conceptual drawing of a vane stress sensor, 



SCOPE 

The primary objective of this research was to determine whether 
the vane stress sensor method of analysis is in fact an accurate and 
economical way for stresses in soils to be measured. This objective 
was pursued in three phases. 

The first phase was a literature review and was separated into two 
sections. Initially, the pertinent references concerned methods of 
measuring in situ lateral stresses in soils. This section of the 
literature review is basically a state-of-the-art summary. The second 
section identifies and reviews critical variables that can be expected 
to affect the preliminary design of the vane stress sensor. Both 
sections of the literature review were a basis for developing an 
experimental program. 

The second phase consisted of laboratory testing. The prime con- 
cerns of this research phase were to answer questions in specific areas 
that could influence design of the vane stress sensor (VSS) . Some of 
these areas are as follows: 

1) Estimate the amount of soil disturbance from insertion of 
blade. 

2) Find the optimum angle of the leading edge on the blade. 

3) Evaluate the effects of blade thickness and surface roughness. 

4) Develop a stress cell that could be used on the VSS. 

The third phase was the development of a prototype for laboratory and 
field evaluation. 



LITERATURE REVIEW 



Determination of Stresses in a Soil Mass 



The existing in situ lateral pressure in a soil mass is called the 
"at-rest earth pressure." This pressure is bounded on the low side by 
the active pressure and on the high side by the passive pressure of the 
soil, where the minimum value or active pressure is reached as a result 
of lateral expansion of the soil prior to failure, and the maximum value 
or passive pressure is reached as a result of lateral compression of the 
soil prior to failure (36) . Horizontal stress for the at-rest condition 
is usually expressed in terms of the vertical stress and the coefficient, 



K : 
o 



a ' = K a ' (1) 

h o v 

where K is the coefficient of at-rest earth pressure, and c^' and c^' 

o 
are the effective horizontal and vertical stresses, respectively (19). 

Accurate determination of the at-rest earth pressure has received 

increasing attention in geotechnical engineering for several reasons. 

Analytical techniques such as the finite element method can now handle 

in situ stresses (20) in making earth pressure predictions. An accurate 

representation of the stress path in triaxial and plane strain testing 

requires that specimens be reconsolidated to the initial in situ stress 

conditions for valid results (32). Other uses of the coefficient of 



10 



at-rest earth pressure (K ) include the design of retaining walls to 
minimize settlement of adjacent buildings, and to predict friction on 
piles. The in situ lateral earth pressure is also important in tunnels 
and bracing systems, where it is necessary to know the initial stress 
in order to determine how much change in stress may be allowed before 
failure will occur in the soil mass. 



Theoretical Determinations of K 

o 



The theoretical analysis of the coefficient of at-rest earth pressure 
has been limited due to problems in modeling real soil behavior. For an 
ideal linearly elastic isotropic material, the coefficient of at-rest 
earth pressure (K ) is related to Poisson's ratio (v) as follows (1): 



K = rr- (2) 

o 1-v 

Wroth (42) developed a theoretical equation for K for overconsoli- 
dated soils that have been unloaded only once. The derivation assumes 
that the stress path upon unloading remains linear, and gives: 

v' 

K = K (OCR) — (OCR-1) (3) 

o nc , . 

1-v' 

where K is the coefficient of earth pressure at rest during normal 
nc 

consolidation of the soil, and OCR is the overconsolidation ratio for 
the soil. The overconsolidation ratio (OCR) is defined as the ratio 
of the maximum vertical stress experienced by the soil to the present 
vertical stress on the soil mass. 



11 



For a soil deposit that has been subjected to more than one cycle 
of loading and unloading due to the deposition and erosion of overlying 
materials, the existing state of stresses in the soil cannot be 
accurately predicted through a theoretical analysis (42) . 
Laboratory Determinations 

The direct measurement of Kq in the laboratory requires a testing 
apparatus in which the soil specimen being loaded can deflect in the 
vertical direction, but is prevented from straining in the horizontal 
or radial direction. Also, during testing, vertical shear stresses caused 
by friction along the sides of the soil specimen must not be allowed to 
develop (7). Such methods of testing have been found to be very time- 
consuming and expensive. Also, "undisturbed" samples are required, 
which are impossible to obtain since stress relief and physico-chemical 
changes occur during sampling (32) . Therefore, a number of researchers 
have correlated Kq with other soil parameters. 

For normally consolidated soils only, several researchers have pro- 
posed the following relationships between Kq and the angle of internal 
friction ( cj>'), of the soil (1): 



Kq = 1 - sin (J) ! (Jaky) (4a) 



Kq = 0.9 (1 - sin (j> ') (Frazer) (4b) 



12 



K q = (1 + 2/3 sin (J)') \ + g|g | (Kezdi) (4c) 



K = 0.95 - sin <j>' (Brooker and Ireland) (4d) 



Brooker and Ireland (10) consider Jaky's equation as an accurate 
prediction of K for normally consolidated sands, whereas their 
equation should be used for cohesive soils. 

Another important factor influencing K is the horizontal stress 
which may be induced as a result of previous loading. Thus, K should 
be proportional to the overconsolidation ratio. As shown in Figure 5, 
a high overconsolidation ratio will dictate a high value for K regard- 
less of soil type. 

A different approach suggested by Spangler and Handy (29) is to 
attribute K to intergranular sliding friction. A rearrangement of 

their equation gives: 

f 

1 - sin (J> 

K = i ■ • — ^ (5) 

o 1 + sin (p 
s 

where <p is the angle of sliding friction, normally ranging from 10 
to 30 degrees. This equation ignores interlocking effects which add 
to (J)', their argument being that volume change work or dilatancy is 
needed for interlocking effects to occur. 

If K is dependent on just the material properties of a soil, 
changes in its value can occur only when there is a subsequent change 



13 



(Jl 



3.0 



2. 5 



£ 2.0 

UJ 



on 

c. 



fe 1.0 



S 0.5 

Li- 
L- ' 
O 

o 




10 20 30 40 50 60 

PLASTICITY INDEX, I 



70 80 



c K as a function of overconsolidation ratio (from 
FlgUre5 ' looker and Ireland (10)). 



14 



in material properties. This means that K will stay constant through 

the entire depth of a uniform soil deposit. But an investigation by 

Massarsch, et al. (20) indicated considerable variations of K close 

o 

to the dry crust, and the general decrease of K with depth below a 
level of 5 meters even though there was no material change. These 
variations may be caused in part by the seasonal fluctuations of the 
water table. Therefore, it can be concluded that no rigorous method 
has been developed that will determine K through analytical or labora- 
tory methods for soils. 
In Situ Evaluation of K 



In recent years attention has been focused on the evaluation of K 

o 

by measuring the total horizontal stress in situ. Basically, the 
techniques available for such measurement can be divided into two 
general classes. 

(1) Hydraulic fracturing; and 

(2) Measurement of the horizontal total stress with a stress cell. 
The hydraulic fracturing method of measuring horizontal stresses in 

clay deposits was suggested by Bjerrum, et al. (8), while they were 
investigating the problem of in situ permeability tests. The hydraulic 
fracturing technique has been described by Massarsch, et al. (20). It 
consists of inserting a piezometer into a cohesive soil, and after 
allowing the excess pore pressures to dissipate, forcing water into the 
soil and causing a crack to form in soil at the piezometer tip. This 



15 



crack is assumed to develop vertically, which means that the hydraulic 
fracturing method is limited to soils with a K value less than 1.0. 
A falling head permeability test is then performed with the quantity 
of water flow being monitored. As the head decreases, a sudden re- 
duction in flow will occur which indicates closure of the crack. The 
head at which the crack closes should be related to the lateral stress 
and can usually be found by plotting piezometer pressure versus flow. 

The assumptions made in the hydraulic fracture method have been 
described by Massarsch, et al. (20) and are summarized as follows: 

(1) The direction of the crack is vertical and the direction of 
the minor principal stress is horizontal. 

(2) The direction of cracking is controlled only by the horizontal 
stress. Factors such as varves , fissures, pockets of highly permeable 
materials, cementation, etc., that could affect the direction of cracking 
are ignored. 

(3) The small, but possibly significant tensile strength of the 
clay is neglected. 

Another shortcoming of this method would be its usage in metastable 
soils, such as loess. These soils are known to collapse upon saturation, 
which would undoubtedly cause erroneous results. Therefore, the 
hydraulic fracturing method would not be applicable in such soils. 

Two procedures have been developed to measure the total horizontal 
stresses in soils using stress cells. The first method, proposed by 
Massarsch (19), is limited to usage in low shear strength material. 



16 



This method consists of pushing a Glbtzl pressure cell, a spade-like 
very thin cell filled with oil, into the soil mass. By taking pressure 
readings from the cell, a plot of horizontal stress versus time can be 
constructed. After approximately one week, the excess pore pressure 
generated by the insertion of the cell will dissipate, and eventually 
an "equilibrium" pressure will be reached. The equilibrium pressure 
is assumed to be the same as the initial in situ pressure. 

Another approach used by Marchetti (43) is similar except that 
the pressure cell is expanded to give a measure of deformation modulus. 
An empirical relationship is established between the horizontal stress 
index Kq and K . On the basis of empirical data fits, K Q = (0.67 K D ) -0.6 
for noncemented and insensitive clays and noncohesive sands, and 
the overconsolidation ratio OCR = (0.5 K D ) " for uncemented cohesive 
soils only. The test thus requires empirical correlations for particular 
soils. 

The other method that uses total stress cells was developed 
independently by Baguelin, et al. (3) and by Wroth (42) . Their methods 
use self-boring cylinders with total stress sensors mounted on the 
exterior. The Wroth device is shown in Figure 6. Since the apparatuses 
are self-boring, the disturbance caused by the introduction of the 



17 



LOAD J 
CELL > 



CUTTINGS 
HEAD 



SOFT w»TER- 
"PROCF FILLED 




DISTANCE - v.-ii 

CRITICAL FUNCTION ' '. " 

O GROUND STRlNCTM *.'j 



TAPER <IO* 



Figure 6. "Camkometer" self -boring pressuremeter (from Wroth (42)) 



18 



apparatus into the soil mass is assumed to be negligible. 

Baguelin, et al. (4) have studied the effect of disturbance 
caused by insertion of their self-boring apparatus. They concluded 
that degree of disturbance does not affect the equilibrium values of 
lateral stress that are eventually obtained, but considerable time 
may be necessary before "equilibrium" is reached and the in situ 
lateral stress is measured. In a recent state-of-the-art paper, 
Wroth (42) stated that under favorable conditions, the radial displacements 
to be expected by insertion of the self-boring apparatus is less than 
0.5 percent of the radius of the cylinder. Ideally, no disturbance 
should occur, but this condition is impossible to achieve since 
shearing forces will always occur on the exterior of the self-boring 
cylinder or, alternately, the borehole must be larger than the probe. 

The self-boring apparatus has been used in both soft and stiff 
soils with reasonable success (42) . The usage in stiff soils gives 
the self -boring method a distinct advantage over the Glotzl cell, 
which can be inserted only in soft soils. A major limitation of the 
apparatus is gravel or rock fragments, which when encountered will cause 
the boring apparatus to jam. 

Although the self-boring apparatus seems to be one of the most 
promising methods of determining the state of stress in a soil mass, 
it still creates a disturbed zone around the sensors. The distur- 
bance has been found to be quite significant, particularly in a 
sensitive highly plastic clay (21). Therefore, although the self 



19 



boring apparatus is perhaps the best method now available, it does not 
give a final answer to the problem of determining the in situ state of 
the stresses in soil (31) . 

Penetration Mechanics 

Disturbance 

The insertion of a vane by pushing into soil will cause a shear 
zone to develop around the vane blades. This zone may be similar to 
one of the three types of shear failure that have been proposed for 
pile foundations. The three modes of shear failure described in the 
literature are: general shear failure (Caquot,and Terzaghi) ; local 
shear failure (Terzaghi, DeBeer, and Vesic) ; and punching shear failure 
(DeBeer, and Vesic) [37]. 

General shear failure is characterized by the existence of a well- 
defined failure pattern that starts downward at the tip of a pile and 
bends outward and upward to the ground surface (40) . This type of 
shear failure can occur only in piles that are relatively short and 
near the surface. Therefore a general shear failure would not be 
expected to be a common failure mode for a vane stress sensor inserted 
to depth-to-thickness ratios greater than 30. 

Local shear failure as described by Vesic (40) is characterized 
by a clearly defined failure pattern that develops only immediately 
below the pile. The pattern consists of a wedge and slip surfaces, 
which start at the bottom of the pile just as in the case of general 



20 



shear. Because of the compressibility of soil, the slip surfaces 
dissipate in the soil mass and do not reach the surface. 

The punching shear failure pattern is not so easily recognized 
(40) , but may dominate deep pile shear behavior. An investigation 
by Vesic (37) indicates that in sands, only punching shear occurs for 
rectangular piles when a depth-to-base ratio of eight is exceeded. 
Through his investigation Figure 7 was developed, which indicates 
the types of failure that can be expected to develop at different 
relative depths. It has been concluded that pile in sands for the 
most part give a punching failure (15) . Therefore a punching shear 
failure probably develops around the blades on the VSS provided the 
blades are inserted into a soil mass past a critical depth. 

The meaning of various failure modes can be seen by reference to 
Figure 8. The general shear plastic zone proposed in Figure 8a is 
probably incorrect because of the assumption that the soil above the 
failure surface A-B, can be replaced by a surcharged loading that does 
not have any strength. The general shear pattern shown in Figure 8b 
requires that the plastic zone extend from the point of initiation 
beneath the pile, through a radial zone away from the pile, and then 
move back upward toward the pile's exterior surface. Local shear 
would be a partial initiation of either of these failure modes, Figure 
8c. Vesic (37) found in his investigation on pile behavior that sand 
did not move through a radial zone and return to the pile surface, and 
Baligh and Scott (6) made the same observation in clay. 



21 



RELATIVE DENSITY OF SAND, D 



.5 



CQ 



S3 

o 

M 

< 

Q 

D 
O 

fa 
O 

ffi 
Eh 
CM 
W 
Q 

W 
> 
H 
Eh 
<C 
vA 
W 

PES 




CIRCULAR 
FOUNDATIONS 



LONG 

RECTANGULAR 

FOUNDATIONS 



Figure 7. Foundation failure modes (from Vesic (37). 



22 




(a) Prandtl 
Reissner 
Caquot 
Buisman 
Terzaghi 




(a) 



(b) 



f -a, 




(c) Berezar.tsev *nd Yaroshenko 
Vesic 



(O 



Figure 8- Assumed plastic zones under deep foundations 
(from Vesic (39)). 

\ 



23 



Punching shear is nicely illustrated by the study by Robinsky, 
et al. (27) on model friction piles in sand. The general displace- 
ment envelope was described as an elongated bulb, and the sand 
revealed a complex variation in density illustrated in Figure 9. The 
zone of displacement beneath the pile point has approximately the 
shape of a truncated cone extending downward and outward from the tip. 
This cone-shaped, vertical compression and two-dimensional horizontal 
expansion were found to take place accompanied by radial downward 
translation. As the pile is advanced and moves through the previously 
compacted zone, sand movement is induced downward along portions of 
the pile wall, causing a sleeve of loose sand to form around the pile. 

These studies found that the process of sand displacement and 
compaction resulted in a seemingly erratic pattern of high and low 
densities which apparently cause a complex system of arching to 
occur. Kezdi (15), and Sanglerat (28) have observed the same type of 
arching effects. It is not possible to predict what effect arching 
will have on the VSS from these studies, but an awareness of this 
phenonenon possibly developing is important. 

Numerous investigations have been conducted to determine the 
radial extent of disturbance caused by the insertion of a pile. 
Vesic (39) developed a theoretical analysis that indicated the radius 
of the plastic zone is dependent primarily on the rigidity index of 
the soil. The size of the disturbed radial zone has been studied by 
numerous investigators and can be expressed as a function of the pile 



24 




Figure 9. Sand displacement around a pile (from Robinsky 
and Morrison (26)). 



25 



diameter. Meyerhof (23) experimentally found that the diameter of the 
disturbed zone in sand is approximately six times the diameter of the 
pile. A pile pushed into saturated clay will cause remolding at a dis- 
tance of roughly one pile diameter (25) . 

The following conclusions relating to the proposed vane stress 
sensor can be made from the preceding discussion on disturbance caused 
by a pile being advanced in a soil mass: 

(1) Inserting a VSS into a soil mass will probably cause a punch- 
ing shear pattern to develop around the vane blades, provided the VSS 
is inserted to a depth greater than approximately 10 times the blade 
thickness. 

(2) Arching in sand around the VSS might cause some problems in 
measuring the in situ stresses. 

(3) The horizontal extent of disturbance out from the surface of 
the VSS caused by insertion into a soil mass should be less than 6 times 
the thickness of the blades and will vary depending on the soil. 

Wedge Shape 

The force necessary to insert the Vane Stress Sensor can be sub- 
stantially decreased by a wedge-shape leading edge, which also should 
influence the degree of disturbance. Therefore, the effect of the 
wedge on the shear pattern that develops over the vane surface must 
be investigated. The two wedge parameters that will probably affect 
the shear zone, are the wedge angle and whether the wedge is smooth or 
rough. 



26 



Baligh and Scott (6) did a study on the insertion of wedges of 
various angles into modeling clay. They found that if the coefficient 
of friction on the wedge surface is larger than a critical value, 
the wedges are considered rough. This value was found to be so low 
that unless extraordinary efforts are made, all wedges can be assumed 
to be rough. 

Meyerhof studied the effect on the bearing capacity of shallow 
foundations by varying the wedge angle. These results indicated for 
wedge angles equal to or greater than 6CP , the bearing capacity factors 
were not affected by changes in wedge angle. When approximately this 
same wedge angle was exceeded in Baligh and Scott's studies with clay, 
a dead zone was found to be about 75° . Therefore, use of a trial wedge 
angle of approximately 6CP for the VSS would appear desirable since 
the resistance to penetration should not appreciably change for 
greater wedge angles. 
Pore Pressure 

Excess pore pressure will develop due to insertion of the vane 
stress sensor. Overconsolidated soils generally develop negative 
excess pore pressure when sheared, whereas positive pore pressure will 
develop in normally consolidated soils (6, 41). 

If a plot is made of the measured total stress as a function of 
blade thickness for an overconsolidated soil immediately after 



27 



insertion of the vane stress sensor, the plot could resemble Figure 10. 

Such a plot should be indicative of negative pore pressure developing 

as a result of dilatant shear, or positive pore pressure as a result 

of soil compaction. It therefore would be advantageous to mount a 

piezometer on the device to measure pore pressures and isolate this 

effect from the total disturbance effects hypothesized in Figure 3. 

A pore pressure device was developed for use at the tip of a cone 

penetrometer by Wissa, et al . (41). 

Stress Cell Locatio n 

The total bearing capacity (Q t ) of a pile (12) is usually 

expressed in terms of skin friction (0 S ) acting on the pile, and point 

resistance (£■ ) : 
P 

Q t = %> + ^s (6) 

The development of skin friction on a pile is dependent on two 
variables, and can be expressed as follows: 

Q s - o" tan <(> s (7) 

where cr' represents the effective normal stress on the pile exerted 
by the soil, and <j> s is the angle of skin friction. The lateral stress 
acting on a pile is not uniformly distributed over its length. There- 
fore, a study of this distribution can give some insight as to where 
the stress cell should be mounted on the vane stress sensor. 



28 



CO 
CO 

PL] 

E-< 
CO 

Q 

H 

CO 

< 

W 

S 





,^"^ 








^ A 


A ^ 




S" 




< \^« 






• ^^^-^_ 




^^---^» V, 




^^"■^So-, 




^*«i*,„ 


~ 


III!, 



FUNCTION OF BLADE THICKNESS 



Figure 10. Representation of possible pore pressure 
effects in soil. 



29 



Numerous investigations have been performed to evaluate the soil 
properties that affect stress distribution on a pile. Vesic (39) 
found that the principle parameters affecting the lateral pressure 
on a pile are strength and volume change characteristics of the soil 
expressed by the rigidity index. Broms and Silberman (9) found that 
relative density of the sand has a large effect on the skin friction 
that develops. As yet, however, the exact distribution of the lateral 
stress cannot be predicted from soil properties alone. 

The lateral stress in sands does not continue to increase indefin- 
itely with depth, but instead reaches a constant value (37) at a depth 
of approximately 15 times the pile diameter. The results indicate 
that the stress cells on the VSS should be installed at a distance of 
15 blade thicknesses below the top of the vane. By doing this the 
reproducibility of the results should be enhanced. 



30 



INVESTIGATIVE PROCEDURE 

The Vane Stress Sensor method of analysis is based on the premise 
that extrapolation to a zero blade thickness effect is possible. If 
this is not the case, the feasibility of the vane stress sensor is 
questionable. Therefore a laboratory testing procedure was set up to 
test this premise. 

Material Properties 

There are two extreme soil types — noncohesive soil such as a sand, 
and cohesive soil such as a homogeneous clay. Both were used in the 
initial research. 
Sand 

The sand initially selected to represent a noncohesive soil was a 
washed well-graded river sand. The gradation of the sand is shown in 
Figure 10. The uniformity coefficient was determined to be 6.25, and 
the specific gravity of the material was 2.66. 

The maximum and minimum densities of the sand were determined by 
following ASTM (1) test procedure D2049-69 , with the following modifi- 
cations. Instead of using a mold 15 cm in diameter by 15.5 cm in height, 
a standard Proctor mold, which is 10 cm in diameter and 11.6 cm high, 
was used to determine minimum relative density of the sand. The maximum 
relative density was determined by using a mold 10 cm in diameter with 



31 









1 




i/i 


IO0 
»0 
10 

to 

«0 

m 

*~ 50 

• 

s*o 
:*° 

; to 

10 


Cutil Sun 


ie-« Slin 


jut 


)rnl 


Cli) Slid 


OIL MECHANICS LABORATORY 

GRAIN SIZE CURVE_ 


Cum |w»«iv~ 1 r i»i 


C«Bf It 1 **< 


41,- I r.. t 


J 


5«m Smi 
' fifi'l" +* *« *>o *?o '•o'to-.oo'joo 




IT 


.Ii... 




-P 


-Ijlll: 




- 


• 




-- 


- 












Ifi 




i 1 


1 


- 






ilii 


Ill 


i 














1 


















-- 


\ 






— 


-- 


— 








— 


li] 






- 


-! 


J 
1 








\ 





















\ 












" 








; 'ii 






ii 


i 








i 






I 


















!! 
















V 










1 








!:(} 


- 


- 




1 


|- 
























\ 










J 








II 
























\ 










1 








ii'l 


i 






1! 


























\ 


















!!i| 






! Ii 

























1 


\ 








■" 








lifl 


- 






[■ 






















Y 












ii 






















\ 








-■ 




_; 














1 


\ l 


























M 










,!i!| 






l| 


3 no > > '* 




















i\ 






1 








111 


f 






I! 




























l\ 
















!i 












= f.' '■■ 


- 






















\ 












-141 
"I 








!i 


— 








o 


1 








































II 








w- 






1 


hi 




L 


- 


1 
1 


"T" 
1 


H 


r> 


- 


1 1 


! 









ju! 

mi 








I 


... 








II it' 






i 








l( 


<c 


10 10 01 

Citu »U« — MitllmtlMI 
TH-.t IN-SITU 5TB[SS RESf*=CH MODEL TEST "ATfS.IAL - SAND 


01 001 0000 


no 


Minis 


■j« llll! 1.. r»n. II t 1 


o r.c.r. 




^. HZ 5 -- 


u 


rUilaua Rcla 


t l»c Dtnfllf 1 


ii.i r.c.r. 










Specific Cr.vltT l.lb 






■ •It • 


M IT t'fl> SIM S(IK 











Figure 11 Grain size curve of sand used in this project. 



32 



P = (psi) 








20 




40 




60 






80 








1 

40% D Sand 
r 




1 




1 






1 


o 






^ Triaxial 























O K-test 




















300 










o 


o 


AO 






— 


200 


" 




4 
O 


o 














100 


O 
o * 

o 

1 


o 

1 


1 




1 




1 









100 200 



300 400 
Oi + a 3 



P = 



, KPa 



500 600 



60 



- 40 



•H 

a. 



_ 20 



PM 



300 



200 



100 



20 



P = (psi) 
40 60 



80 




P. 



P = 



, KPa 



Figure 12. Shear strength test results for sand. 
1 psi = 70.31 gm/sq cm 



33 



the sand being compacted to a final height of 10 cm + .25 cm. The 
maximum density of the sand was found to be 2.00 g/cc and the minimum 
density was 1.76 g/cc. These values, as was expected, are close to 
one another because of the gradation of the sand. 

Maximum and minimum relative densities normally are not encountered 
in a natural deposit. Therefore, all testing was done on sand compacted 
to either 40% or 75% relative density, D . 

The shear strengths of the sand at both testing densities were 
determined by two different testing methods. The first method consists 
of running a set of three triaxial tests (consolidated drained) on sand 
samples of each density. Figure 12 shows the results of these tests. 
It should be noted that this is not a graph of normal stress versus 
shearing stress, but rather a stress path method (17) of plotting test 
results. A functional relationship exists between the internal angle 
of friction and the slope of the best-fit line through the triaxial points 
on the p-q plot. This best-fit line was determined through a least 
square fitting method, and the resulting internal friction angles were 
calculated. The dense sand had a friction angle of 41° with a corre- 
lation coefficient of 0.998, while the loose sand had a friction angle 
of 39.5° with a correlation of 0.994. 

The other shear strength testing method is the Iowa K-Test (18) , 
diagramed in Figure 13. This apparatus consists of a split cylindrical mold 
with a piece of Teflon covering the split. When a soil sample is placed in 
the mold and loaded in the vertical direction, it begins to deform 
in both the vertical and radial directions. 



34 




TEFLON SEAL 



K-TEST 
MOLD- 



i 



K5 



DIAL GAUGE 



Figure 13. The Iowa K-Test mold (from Lutenegger (18)) 



35 



This causes expansion of the mold which can be monitored with a 
dial gauge. Since the relationship between radial stress and mold 
deflection has been previously determined, it is possible to determine 
the radial stress acting on the soil sample at predetermined vertical 
stresses. The sample is considered to always be in shear, so the shear 
strength parameters of the sample can be determined through Mohr-Coulomb 
failure criteria. 

K-Tests were performed on sand samples in both the dense and loose 
state. The results of these tests are plotted on the p-q graphs in 
Figure 12 with the radial and top platen stresses considered to be 
principal stresses. The angle of internal friction for the dense sand 
was 41.6° with a correlation coefficient of 0.999, and the loose sand 
had a friction angle of 39.0° and a correlation coefficient of 0.999. 
These values are in close agreement with triaxial test results, as 
may be seen in Figure 12. 

The soil-to-steel friction is another property that can be deter- 
mined from the K-Test. This is done by setting a soil specimen to be 
tested on a pressure cell. A difference in stress is noted between 
the stress at the top of the specimen and the bottom of the specimen. 
This difference in stress is caused by side friction occurring between 
the mold and soil specimen. It is assumed that a uniform gradient of 
friction exists across this interaction surface. By assuming this 
gradient and knowing the stress (radial stress) acting normal to this 
interface, it is possible to calculate the friction angle of soil on 
steel. 



36 



K-Test results on a dense sand gave a friction angle of soil on 
steel equal to 17.5° with a correlation coefficient of 0.984 (10 data 
points). The loose sand had a soil to steel friction angle of 16.5° 
with a correlation coefficient of 0.994 (10 data points used). 
Modeling clay 

Modeling clay was used by Baligh and Scott (6) in their laboratory 
work on wedge penetration into clay for the following reasons: it has 
high cohesive strength, normally is homogeneous, and approaches rigid- 
perfectly plastic behavior. These save- characteristics were desirable 
in this research, so a modeling clay manufactured under the commercial 
name Roma Plastilina was selected to represent a clay soil. The model- 
ing clay was available in two colors, hite or gray, but it was avail- 
able in only one consistency. 

The bulk density of the clay was determined by the following 
method of testing: the clay was compacted into a standard Proctor mold 
(944 cc) in three lifts. Each lift received 25 blows from a 4.5 Kg 
hammer dropped 46 cm. A Jensity of 1.55 g/cc was determined through 
this method, and all subsequent testing was done on modeling clay 
molded in this same way. 

In order to evaluate the effect of different cohesive material 
strengths on the ability of the vane stress sensor to perform properly, 
it became necessary to change the consistency of the commercially 
available modeling clay. This was done by adding mineral oil (0.7% by weight) 



37 



to the clay. This increased the cbhesive strength of the clay 
and allowed testing on modeling clay with two different consistencies. 
X-ray diffraction technique was used to determine the modeling 
clay mineralogy. A copper x-ray tube was used in the analysis with a 
nickel filter. The x-ray trace of the clay is shown in Figure 14. Trace 
A was run on a sample of untreated clays. By observing the 20 angles at 
which peaks occur, it was possible to identify the presence of kaolinite 
(11) in the clay sample. But, at the lower 20 angles peaks were observed 
that were not characteristic of kaolinite. A second x-ray trace was made 
on a clay sample that had been heated to a temperature of 400°F for four 
hours to remove organic matter. The trace for this sample, Figure 14, 
indicates the low 20 peaks disappeared and the clay is composed of one 
clay mineral, kaolinite. The clay in its commercial form also contains 
some organic compounds . 

Shearing strength of the material was determined by the same 
methods of testing used for sand. A set of three triaxial tests 
(unconsolidated undrained) were performed on the softer commercial 
form of modeling clay. The test indicated an internal friction angle 
of 4° and 3 psi (21 KPa) cohesion, the correlation coefficient being 
0.998 for this data. The K Test also gave these same values for c|> and 
c, with a correlation coefficient of 0.961. A K Test of the stiff 
clay gave a friction angle of 8° and a cohesion of 4 psi (28 KBa), 
with a correlation coefficient of 0.998. 



38 




] 






e 
>h 

0) 

o 

s 
o 

01 

(U 

O 
cfl 
!-< 
■U 

d 
o 

•H 
■U 

a 
cd 

u 

•H 

>> 
CO 
^ 

I 
X 



M 

3 

•H 



J s 



39 



The soil-to-steel friction parameters were calculated for 
each modeling clay consistence in the same way as for the sand tests. 
Results indicated an adhesion term of soil-on-steel was present in 
the modeling clay test, the friction angles being 6° and 5° with an 
adhesion of 3 KPa and 2.5 KPa respectively for the stiff and soft 
clay. Both had correlation coefficients greater than 0.94. 

Blade Insertion Into Sand 

The purpose of laboratory testing was to determine the amount of 
disturbance and resulting stress increase caused by the insertion of 
a rigid steel blade into a soil mass. The ideal soil mass would be 
semi-inf inite, but this of course is impossible. Two possibilities 
were available that could approximate a semi-inf inite mass, one being 
a box of large dimensions, and the other being an Iowa K-Test mold. 

The box had two major shortcomings in this type of testing. If 
the sides of the box are rigid a boundary condition is inevitable. The 
other problem is the development of friction forces along the sides 
of the box when vertical or lateral loads are applied to the soil mass. 

Selection of an Iowa K-Test mold substitutes an elastic yielding 
boundary but with an inherent high modulus of elasticity compared to 
that of soils. This high modulus should cause higher stress changes 



40 



to occur when a rigid inclusion is inserted into a soil specimen 
contained in the mold. Therefore usage of a K-Test mold should 
exaggerate the error and be a conservative step. The main advantage 
of using a K-Test mold instead of a box is that the mold expands in 
the lateral direction, more-or-less simulating confinement in a semi- 
infinite mass. Therefore, the K-Test mold was selected for laboratory 
testing. A second advantage is that the lateral stress at all times 
may be monitored from expansion of the mold, allowing a ready evalua- 
tion of the stress effects from blade insertion. 

The rigid inclusions used to represent a vane stress sensor are 
steel blades 6 inches long, one or two inches wide, and 1/8 in., 3/16 
in. , or 1/4 in. thick. The blades were tapered at one end by an apical 
angle of 45°, 60°, or 90°. The surface of each blade was either ground 
smooth or sanded rough with 60 grit emery cloth. No attempt was made 
to use scaling laws in the design of these blades. Previous researchers 
(38) have found that conventional dimensional analysis of model tests in 
sand fail. Scale effects have to be assessed in a different way, because 
they relate to changes in intrinsic properties of the material. 

The testing procedure used is outlined in the following steps: 
1. The K-Test mold was expanded by inserting a soft rubber 
specimen and compressing it under load. When the mold reached a 
specific lateral deflection, a metal bar was slipped into the mold slit. 
Then the rubber specimen was unloaded allowing the K-Test mold to 
tighten on the metal bar, which kept the mold from reaching its initial 



41 



unstressed state. Therefore a certain known stress is locked into 
the mold. The rubber specimen is then removed. 

2. The mold is placed on a pressure cell, as shown in Figure 15. 
Sand is poured into the mold and compacted to a predetermined relative 
density, either 40% or 75%, by tapping the mold. All sand tests had 
final sample heights of 4 in. 

3. The mold and pressure cell are transferred to a loading 
machine. A stationary Teflon disc was clamped on top of the sand 
(Figure 15). The metal bar in the mold split is removed, causing a 

lateral and vertical pressure to be applied to the sand, since it is 
completely enclosed in the mold. The lateral reading of the mold is 
noted and is used in calculating the initial lateral stress. Also, 
the base pressure gauge reading was recorded. 

4. It was initially assumed that blade penetration rate would 

not be a significant factor in sand test. Therefore a seemingly reason- 
able value of 0.0275 cm/sec was selected for test blade insertion in 
all sand tests. At blade depth penetration increments of 0.25 in. 
blade load, mold lateral deflection, and base pressure were recorded 
until the test blade reached a depth of penetration equal to 3.5 in. 

5. The blade was then extracted at a rate of 0.0275 cm/sec. The 
same readings were taken in this step as in step 4, and at the same 
depths. This was done until the blade was completely extracted. Then 
the sample was extruded from the mold. 

This laboratory testing procedure was modified when special horizontal 
layer samples were made to investigate the amount of disturbance caused by 



42 



Elastically 
Yielding 
K-Test 
Mold 



Rigid Clamp 



Teflon Disc 




Blade Pressure Gauge 



Figure 15. Laboratory testing apparatus. 



43 



blade insertion. Sand layered samples were constructed in the K-Test mold 
with sand containing 1% by weight portland cement. Initially a natural colored 
sand layer approximately 0.5 in. in height was placed in the mold. 
Then a 1/4 in. thick layer of sand with a small amount of lamp black 
mixed into it was placed in the mold. This stacking procedure was 
followed until a sample 4 in. high had been constructed. 

After the blade had been removed from the layer samples, the 
samples were saturated with water and allowed to set 24 hours. At the 
end of this period the sample was extruded from the K-Test mold and 
sliced at predetermined vertical planes. Each sample was then photo- 
graphed so that measurements on amounts of disturbance could be 
recorded. 
Data reduction 

Two methods for data reduction were available from blade insertion 

tests in the K Test mold. The first, based on the blade skin friction, 

uses the following equation to obtain an average lateral stress on the 

blade: 

F 
% = A; tan <D s (8) 

where 

a = average lateral stress on blade 

cj) = angle of friction for soil-steel interface (from K-Test data) 



44 



*b 



= area of blade in soil 



F = pulling force on blade. 

During insertion of the blade, the load acting on it is composed 
of F plus a force F exerted against the bottom of the blade. While 
pulling the blade, F is equal to the total force acting on the blade. 
The calculated F for each case, pushing or pulling, have been found 
to be unequal by numerous investigators (9, 35, 36). This is probably 
because downward friction from the blade adds to the overburden 
pressure, with a consequent increase in the lateral pressure on the 
blade. Conversely, the friction developing during pulling acts 
opposite to the overburden pressure close to the loaded blade. Due to 
uncertainties in estimating the bottom resisting forces, only blade 
loads recorded during pulling were used in calculating the lateral 
stress on the blade. 

The second method for data reduction was simply to compare the 
K-Test mold stress after blade insertion to the stress before blade 
insertion. 

A problem in data reduction was that in order to define the 
functional relationship between blade thickness and lateral stress, 
all variables except blade thickness should be held constant. It was 
found that the initial lateral stress on different samples varied by 
as much as 2 psi. Therefore, a nondimensional ratio of lateral 
stress on blade or mold to initial lateral stress on sample before 



45 



testing was defined and called X. This allowed plotting of a graph 
with blade thickness on the abscissa and X on the ordinate. An 
extrapolation of data points to a blade of zero thickness ideally 
should give a X of one. 
Blade Angle 

The literature indicated that a blade apical angle of 60° would 
probably minimize disturbance effects caused by blade insertion, but 
it was deemed advisable to perform some laboratory checks before the 
major portion of laboratory testing was undertaken. All of these tests 
were done with smooth-finished blades 2 in. wide and 1/8 in. thick. 
The results are shown below. 



Blade D , Sand 

Angle Density 



45° 40% 

60° 40% 

60° 75% 

90° 75% 



As can be seen from the results, X was minimized when the blade 
had an angle of 60°. This is in agreement with previous researchers, 
and all subsequent sand tests were done on blades with 60° tapered ends. 
Mold effects 

In all testing, data were reduced by assuming a uniform distri- 
bution of stress across the surface of the blade. To test this assumption, 



Blade 


Mold 


14.84 


10.17 


2.12 


2.84 


9.41 


8.65 


14.21 


16.78 



46 



a special blade was fabricated as shown in Figure 16. The dimen- 
sions of the blade were 7 x 2 x 1/8 in. One side of the blade was 
ground down so that six pieces of steel tape covered with a layer 
of shim stock would lay flush on the blade surface and would maintain 
the initial blade thickness of 1/8 in. 

Four pieces of the steel tape had the same dimensions, 6 x 0.25 x 
0.005 in. The other two pieces had a width of 0.5 in. The tapes were 
laid side-by-side against the blade surface, the two wider strips 
being at the outside, and a layer of 0.005 in. shim stock was placed 
over them and epoxy glued to the blade just above the angle taper. 

This blade was inserted into a K-Test mold containing sand at 40% 
relative density. The previously described procedure for blade inser- 
tion was followed; then when the blade reached the final penetration 
depth of 3.5 in., the strips were pulled sequentially a distance of 
approximately 0.050 in. The maximum force encountered for each tape 
was recorded during the pulling operation. 

Later, this same blade was inserted into a small box 2ft. x 1.5ft. x 
4 in. in dimensions filled with sand at 40% relative density. Again 
the strips were pulled with the maximum resisting force on each 
recorded. 

Two tests each were run in the K-Test mold and in the box. The 
results of these tests are shown in Figure 17, with a plot of strip 
location as the abscissa and the ratio of pulling force per inch 4- 
maximum force per inch width as the ordinate. 



47 





EXPOSED FRICTION STRIP 



SHIM COVER STOCK 



Figure 16. Photographs of special stripblade with friction strips, 



48 



Z 
O 
H 
Eh 
U 
H 

> 
H 
Eh 

H 



1.0 



0.5 



A Test 1 
• Test 2 



2.0 



^ 



■A— 

— • — ■ 



-&■ 



2.5 3.0 3.5 4.0 
DISTANCE FROM MOLD SPLIT, in. 



1.0 - 



o 

H 
Eh 
U 
H 
Pi 
fa 

g 0.5 

H 
Eh 
< 

w 



O Test 3 
▲ Test 4 



6.0 



o 



± 



^ 



O 



_L 



6.5 7.0 6.5 6.0 

DISTANCE FROM BOX SIDES, in. 



Figure 17. Test Results from special "tape" blade in sand. 
1 in = 2 . 54 cm 



49 



The top plot in Figure 17 shows the results when the special 
blade was inserted into sand confined by a K-Test mold. In test 1, 
the first tape pulled was the one farthest from the mold split, 
whereas in the second test, the tape closest to the mold split was 
pulled first . By doing this , error introduced by pulling sequence 
would be averaged. The plots for these two tests indicate a non- 
linear stress distribution across the blade. The data points on the 
side of the blade nearest the mold split are closer to being equal in 
value than the other points. This is probably due to the way in 
which the mold opens when the blade is inserted into it, allowing 
more soil movement to occur on the half space closest to the mold 
split causing a more uniform stress distribution to develop. The 
plot of data points from box tests 3 and 4, Figure 17, indicate a 
more uniform distribution of stress across the blade, by avoiding 
assymetrical loading conditions in the K Test mold. 

Blade Insertion Into Modeling Clay 

Modeling clay tests followed the general procedure outlined for 
sand tests, with the exception of sample preparation. The clay was 
not compacted in the K-Test mold. Instead it was compacted in a 
standard Proctor mold, extruded, wrapped in a sheet of aluminum foil 
and slipped into the K-Test mold. Aluminum foil was used to increase 
the overall volume of the specimen, resulting in a higher initial 



50 



v *?. n .r'^' 







k&vi 









*Sf 






3*S£ 















Figure 18. Photograph of a layered modeling clay specimen 
with a metal blade inserted into it. 



51 



lateral stress. 

Some layer samples were constructed with the low consistency clay. 
This was done by compacting two specimens, one gray and one white, 
and slicing them into "biscuits" approximately 0.5 inches in height. 
The biscuits were stacked in alternating colors to a height of 4.5 
inches. Care was taken in this stacking process to make sure the 
samples retain a right circular cylinder shape. 

Figure 18 contains a photograph of a layer clay specimen sliced 
open to expose a 1/8 in. blade that has been inserted into it. 
Normally the blade would be extracted so that the stress acting on 
its surface could be calculated, this specimen being prepared for 
illustrative purposes only. 
Data reduction 

Because the friction between clay and steel was found to be de- 
pendent on two parameters, namely friction angle and adhesion, the 
equation used in clay test data reduction was : 

F - oA, 

5 U = IT -r- (9) 

b A^ tan <j) 



where 



a = average lateral stress on blade 

<J) = angle of friction for soil-steel interface 

a = adhesion for soil-steel interface 



52 



\ 



= area of blade in soil 



F = shear force acting on blade surface 

The definition of the dimensionless term A is the same as in the 
sand tests. 
Rate of Penetration 

Creep and pore pressure are two physical phenomena common in clay, 
and have a time-dependent behavior. Therefore the penetration rate 
at which a blade is inserted into a clay specimen became more important 
than it was during sand testing, and a short investigation was done 
to determine a rate by which changes from the initial stress regime 
would be minimized. 

Previous research work on cone penetration rates indicated that a 
tenfold increase or decrease in rate was necessary to significantly 
change resistance value of penetration in a clay deposit (16) . There- 
fore tests were performed at three penetration rates on the low consis- 
tency clay. A smooth blade 2 in. wide and 1/8 in. thick with a taper 
angle of 60° was used in all three tests. The results are summarized 
below. 



Penetration Rate (cm/sec.) 



Blade 



Mold 



0.00275 

0.0275 

0.275 



6.17 

4.32 

11.15 



2.12 
1.41 
5.5 



53 



The results indicate the rate previously used in sand testing is 
also optimum for clay testing, so this rate was used throughout the 
remaining laboratory tests. 
Blade Angle 

The optimum angle for the blades used in clay testing was deter- 
mined by a procedure similar to that used in sand. Three blades with 
differing angles on their tips were inserted into low consistency 
clay specimens. The blades all had a smooth finish, a width of 2 in. 
and a thickness of 1/8 in. The results of these tests are listed 
below. 



Angle Blade Mold 

45° 0.78 1.94 

60° 4.32 1.41 

90° 6.11 4.58 



The selection of the optimum angle was not as clear-cut as it 
had been in the penetration rate selection. The mold X values indicate 
a 60° angle causes the least stress increase, but a 45° angle produced 
the lowest blade X. If A is calculated correctly, it seems reasonable 
that no X would be less than one, since a lower value would indicate 
a decrease in stress due to the insertion of the blade. This could 
only occur by developing a negative pore pressure. Therefore, because 
of the questionable blade X value calculated for the 45° angle, more 
weight was placed on the mold X values. These values suggest the 60° 
blade is optimum, and this was used in all subsequent clay tests. 



54 



STRESS SENSOR DESIGN 

The pneumatic stress sensor used in this research was devised 
by Dr. R.L. Handy and Dr. E. G. Ferguson and introduced in February 
1977 (13). To evaluate the workability of the proposed sensor, the 
following laboratory investigation was undertaken. The perfected 
sensor then was incorporated into a metal blade for further laboratory 
testing in sand and modeling clay. 

Model Test 

Procedure 

The initial laboratory set-up used in developing the stress cell 
was designed by Dr. Glen Ferguson, then Chief Engineer for Soil 
Systems, Inc. It consists of a needle valve, pressure gauge, sensor 
simulator, air lines, and two consoles. Each console was equipped 
with a pressure regulator, pressure gauge, and air tank. 

The set-up is shown in Figure 19. Console 1 was used to supply 
air at a constant pressure through line 1 (approximately 7 ft in 
length) to a needle valve. The pressure in this line was kept at 
either 60 psi or 80 psi, and is indicative of a "tank pressure." 
The needle valve was used to regulate the air flow through line 2 
(approximately 1.5 ft in length) to the sensor. A pressure 
gauge was placed between line 2 and the needle valve, so that 
calibration of the sensor could be attained. Later, for reasons 



55 



o 
c/f 



o 
o 





CM 



Lu 



UJ , 



c 

I— 

csr <: 

o _l 

^ s: 

UJ i— i 

00 oo 







o 
oo 

2: 
o 

C_3 





UJ UJ _ 

uj < £J 

UJ >• 

2: 


; 


l 


1 











>- 

.-1 

UJ 

_J 







rD uj 

00 o 

CO <t 

UJ <J3 



■u 

CJ 

c 
o 

CJ 

J-l 

o 

CO 

c 

01 
cc 

0) 
S-i 
3 
CO 
CO 

cu 

S-i 

a, 
O 

i-H 

U-l 

I 

CO 
CO 
60 

00 

c 

•H 
•U 
CO 
CU 



T3 

CC 

ex 

D 
I 

AJ 

0) 
co 

u 

O 

ca 
u 
o 

J=> 

CO 
.H 

O 



C3 

CU 

a 

00 



^ 



cu 
u 

3 

•H 



56 



yet to be discussed, a flow meter was placed between the pressure 
gauge and line 2. Console 2 was used to supply a constant air 
pressure to the soil simulator chamber. The consoles were standard 
equipment for the Iowa Borehole Shear Tester manufactured by Handy 
Geotechnical Instruments, Inc., Ames, Iowa. 

The sensor simulator pictured in Figure 20 was designed to give 
maximum flexibility in changing sensor membranes and also to allow 
pressure applied on the sensor to be adjustable. The simulator 
consists of the sensor body, which had two air inlets each of which 
enter into separate inlet chambers. Between the inlet chambers was 
an exhaust chamber that was open to the atmosphere. Covering the 
sensor body was a diaphragm such as shim stock. The diaphragm was 
held firmly against the sensor body with a hold-down disc. An air- 
tight gasket was placed between the hold-down disc and the soil simula- 
tor chamber. The soil simulator chamber was used to simulate the pressure 
of a soil mass. The soil simulator chamber had an air inlet that 
allowed the pressure in it to be varied, consequently the pressure on 
the sensor was also varied. 

The sensor simulator operates in the following manner: 

1. The pressure in the soil simulator chamber is increased and 
held at a predetermined value. This causes the diaphragm to seal 
tightly against the sensor. 

2. Air is forced into the sensor inlets causing air in the inlet 
chambers to increase in pressure. The pressure increase continues until 



57 



AIR INLETS 



INLET CHAMBER 



EXHAUST CHAMBER 



SCREW HOLES 



AIR INLET 




SENSOR 



DIAPHRAGM 
$HTM STOCK) 



DIAPHPAGM 
HOLD-DOWN DISC 



GASKET 



SOIL SIMULATOR 
CHAMBER 



Figure 20. Drawing of sensor simulator used to evaluate 
the pneumatic stress cell 



58 



it just exceeds the soil simulator chamber pressure. At this time the 
diaphragm bulges out away from the sensor. The air in the inlet chamber 
rushes into the exhaust chamber, resulting in a pressure drop in the inlet 
chambers . 

3. The pressure drop will cause the diaphragm to collapse to its 
original position against the sensor body. This will reseal the inlet 
chambers allowing the pressure in them to again increase, thus beginning 
the same cycle of events over again. This cyclic operation of filling 
the inlet chambers with air and then after pressure build up allowing 
them to exhaust, gave the sensor a "fluttering" effect. 

A specific testing procedure was set up in hopes of answering the 
following questions: 

1. Can the gas-flow pressure sensor be taken from concept to a working 
model? 

2. If so, what diaphragm material should be selected for the sensor? 

3. Does the length of air lines have any effect on the sensor 
calibration? 

To answer the first question, 0.002 in. thick brass shim 
stock was placed in the sensor simulator. Console 2 was set at 10 psi, 
thus applying the same pressure onto the sensor by way of the soil 
simulator chamber. At this time console 1 was set at 60 psi, and the 
needle valve slowly opened. The pressure in line 2 was increased until 
it reached a maximum value of approximately 14 psi. At this point the 
sensor started fluttering. The fluttering was indicated by an audible 



59 



hissing, due to gas being exhausted into the atmosphere, and an oscilla- 
tion of the pressure gauge needle. Maximum and minimum readings 
indicating the pressure in line 2 during both bulging and return of the 
diaphragm were taken. 

After the maximum and minimum pressure readings had been recorded, 
the pressure in the soil simulation chamber was increased in increments 
of 10 psi. At each increase, the pressure in the line to the sensor was 
also increased until the sensor again began "fluttering." During this 
time the maximum and minimum gauge pressures were recorded. This process 
continued until the limiting tank pressure was reached. The results 
of this test were plotted with the soil simulator chamber pressure on the 
abscissa and the sensor pressure on the ordinate. Remarkably linear 
plots were observed from these results. Therefore, it was concluded the 
stress sensor concept could be turned into a functional sensor that would 
be easy to calibrate. 

Selection of the best material for the diaphragm was done in a trial- 
and-error process. Basically it involved trying different materials (e.g.. 
steel, brass, and aluminum) and different thicknesses to arrive at the 
presumably best membrane. A plot of soil simulator pressure versus 
sensor pressure was made for each type of shim stock tested. An 

example is shown in Figure 21. 

The effect of varying line lengths was investigated on both line 1 
and line 2. Initially, a twenty-five foot extension was added to line 
1. A new calibration was performed on the sensor, and compared to the 
previous calibration curve. 



60 



-H 
CO 








V 


kb 








A 


Q) 

u 40 






V 




w 

CO 
CD 




X 


A 




^20 










o 

CO 
C 
d) 
CO 


_ X 











I 


i 


1 


1 1 1 I 1 



20 40 60 80 
SOIL SIMULATOR PRESSURE, psi 



100 



60 












tf 






-H 


















CO 












A 






a, 










V 








v 


















CD 










▲ 








u . 


















3 40 


















CO 








V 










CO 








A 










CD 


















u 


















CU 


















^ 






A 












o 20 

CO 
















v MAXIMUM 


c 


















CD 

CO 




2 












A MINIMUM 







i 


1 


i 


1 


1 


J 


l l 1 



20 40 60 80 
Soil Simulator Pressure, psi 



100 



Figure 21. 



Typical plot of results obtained from 
a steel diaphragm performance test 
1 psi = 70.31 gm/sq cm 



61 



This same method was to be used when a 25 ft extension was placed 
on line 2. When line 2 was increased in length, no oscillation of the 
pressure gauge could be observed. The audible fluttering noise of the 
sensor was heard, but the compressibility of the air in the longer line 
caused the previously observed pressure gauge fluctuation to be non- 
existant. This meant that the operator would not know the sensor had 
begun fluttering; instead he probably would continue increasing the 
pressure in line 2, which would only result in an increase in the sensor 
response rate. 

The use of a remotely-located pressure gauge to monitor the pressure 
at which a sensor began oscillating was questioned. One alternative would 
be to measure the pressure with a transducer that was located close to 
the sensor; another was to make an electrical contact between the sensor 
membrane and the sensor body such that when the sensor started fluttering 
a break in the circuit would result. Either of these ideas, if implemented, 
would require electrical lines to be placed "down the hole" when the vane 
stress sensor was used. Also, the electrical circuit breaker idea could 
cause an anode to develop which might result in considerable corrosion 
to the stress sensor. An alternative was devised to solve the fore- 
mentioned problem. 

By reasoning that after each cycle (that is pressure build-up then 
release) a surge of air would take place in line 2, it should be possible 
to detect the "surge" with an air flowmeter. A flowmeter with a range 
of 0-100 ft /hr was placed after the pressure gauge as pictured in 



62 



Figure 20. The addition of the flowmeter made it possible to construct 
a plot similar to Figure 20, even though line 2 had been increased in 
length by 25 ft. 
Preliminary Diaphragm Selection 

Test results on different diaphragm materials were compared by a 
least-square line fitting method applied to data points. A summary of 
test results can be found in Table 1. The constants "c" and "d" represent 
the slope and intercept of the previously mentioned line. Correlation 
coefficients, r, of the lines were used to check the strength of the 
relationship between the two pressures. 

Three materials initially were tried as sensor diaphragms, their 
pertinent properties being listed in Table 2. Aluminum shim stock used 
in the sensor was found to be inadequate because yielding occurred at 
low stresses, making calibration impossible. Both brass and steel shim 
stock at equivalent thickness produced approximately the same constants 
"c" and "d". Steel was tentatively selected as a sensor membrane due to 
its high modulus of elasticity and low coefficient of thermal expansion. 
Also, the sensor membrane is repetitively "fluttered" so a material such 
as steel with a high endurance limit is needed. 

Membrane thickness selection was governed by two criteria: the 
thickness should be maximized for durability, but the maximum thickness 
will be limited by the requirement that the diaphragm be flexible enough 
to properly seal against the sensor body. Three different thicknesses 
of shim stock were tested; results are summarized in Table 1. 



63 



Table 1. Summary of Diaphragm Test Results 

1 in = 2.54 cm , 1 psi = 70-31 gm/sq cm 









Tank 


Gauge 














Thickness 


Pressure 


Reading 










Number 


Material 


(in.) 


(psi) 


(Max/Min) 


c 


c 


r 


Trials 


Bl 


Brass 


.001 


60 


Max. 


1.22 


3.22 


0.999 


2 








60 


Min. 


1.13 


2.32 


0.994 


2 








80 


Max. 


1.23 


2.20 


0.999 


2 








80 


Min. 


1.13 


1.85 


0.998 


2 


B2 


Brass 


.002 


60 


Max. 


1.29 


-0.11 


0.998 


2 








60 


Min. 


1.11 


1.82 


0.998 


2 








80 


Max. 


1.21 


1.50 


0.999 


2 








80 


Min. 


1.07 


2.73 


0.998 


2 


B3 


Brass 


.005 


60 


Max. 


1.20 


2.75 


0.998 


2 








60 


Min. 


1.12 


2.75 


0.998 


2 








80 


Max. 


1.20 


2.45 


0.999 


2 








80 


Min. 


1.06 


3.40 


0.999 


2 


B4 a 


Brass 


.002 


60 


Max. 


1.19 


1.25 


1.000 


1 








60 


Min. 


1.06 


2.75 


1.000 


1 








80 


Max. 


1.19 


1.55 


0.999 


1 








80 


Min. 


1.06 


3.00 


0.999 


1 


B5 b 


Brass 


.002 


60 




1.10 


10.50 


0.997 


1 








80 




1.18 


9.65 


0.998 


1 


SI 


Steel 


.002 


60 


Max. 


1.24 


0.87 


0.999 


2 








60 


Min. 


1.09 


2.00 


0.999 


2 








80 


Max. 


1.24 


0.63 


0.999 


2 








80 


Min. 


1.07 


2.23 


0.998 


2 


S2 


Steel 


.005 


60 


Max. 


1.17 


0.125 


1.000 


2 








60 


Min. 


1.05 


0.375 


0.999 


2 








80 


Max. 


1.16 


0.29 


0.999 


2 








80 


Min. 


1.03 


1.09 


1.000 


2 



a Additional 25 ft. air line added between console 1 and needle valve, 

b Additional 25 ft. air line added between pressure gauge and sensor 
simulator. 



64 



Table 2. Average properties of selected enginnering materials. 
1 in = 2.54 cm, 1 ksi = 70.31 kg/sq cm 

Modulus of Coefficient of Thermal 

Elasticity Expansion Endurance Limit 

Material (1000 ksi) (10 in/in. /F) (ksi) 

Aluminum 10.3 12.5 6-11 

Brass 15 9.8 7-20 

Steel 29 6.6 24-32 



The 0.005 in. shim stock did not seal properly on the 
sensor body, and air leaked from one inlet chamber to the exhaust 
chamber when this thickness was tested. Both the 0.001 in. and 0.002 in. thick- 
nesses appeared to work properly. Therefore it was suggested that the 
thicker of the two (0.002 in.) be used in future sensors. 

In all sensor tests high correlation coefficients were obtained. 
Since the data points used in the least-square line fitting method were 
normally from two independent trials, high correlation coefficients 
should give some indication that calibration of the sensor is highly 
reproducible. 

Line length effects were studied in tests B4 and B5. The B4 test 
had a 25 ft extension on line 1. The effect of this increase in line 
length was fairly insignificant. But the B5 test, which had a 25 ft. 
extension on line 2, showed an increase in the. "d" coefficient (see 
Table 1). This increase, though significant, is probably due to fric- 
tional resistance in line 2 and the air experiencing density changes in 
the line during sensor operation. It should be noted that in this test 



65 



maximum and minimum pressure gauge readings were not recorded. Instead, 
only a single reading could be taken since no oscillation of the pressure 
gauge was observed. Actuation of the sensor in this test was indicated 
by oscillation of the ball in the flowmeter. This method of sensor 
calibration was found to be extremely precise . 

Blade-Mounted Sensors 

Since encouraging results were attained in testing the stress 
sensor concept and in selecting a sensor diaphragm, it was decided to 
continue the study with soil contact rather than air contact. A new 
stress sensor was designed and mounted on a steel blade . The overall 
blade dimensions are 7.0 x 2.0 x 1/8 in. The blade had a smooth 
finish and a 60° angle taper on its bottom. The blade, like all pre- 
viously used blades, was made of cold-rolled steel. 

The sensor was located 3.8 cm from the bottom of the blade on 
center. Intake and exhaust lines were machined into the blade on the 
side opposite the sensor. After machining, these lines and the sensor 
were covered with a layer of shim stock. 

The sensor was made circular in shape to avoid stress concentrations 
at corners. The overall diameter of the sensor was one inch, which is 
equal to one half the blade width. The inner chamber was used as an 
exhaust chamber, while the outer chamber was equivalent to the inlet 
chambers used in the model stress sensor tests. 



66 



Results from the stress sensor development phase of this study 
indicated that steel shim stock 0.002 in. thick would be the optimum material 
for a sensor membrane. Initial attempts were made to cover the stress 
sensor with this grade of shim stock by silver soldering them together. 
Unfortunately, heating the thin piece of steel caused it to develop 
weak spots that probably would oxidize in a short period of time, 
rendering the sensor useless. Therefore, stainless steel was used 
instead of cold-rolled steel, and epoxy-cemented into place. Although 
no testing was done during the preliminary selection tests on stainless 
steel, it was felt that its physical properties were sufficiently close 
to those of cold -rolled steel that the sensor should perform properly. 

One further modification was done to the testing apparatus before 

any testing was done with the sensor blade. The flowmeter previously 

3 
used was replaced with a purge meter that had a range of 0.2 - 3.5 ft /hr. 

to increase the sensitivity of the apparatus. 

The sensor blade was calibrated in a cell filled with water. 

Pressure in the water cell was increased in increments of 5 psi to 50 

psi, and at each increment the pressure into the blade inlet chamber was 

increased until the purge meter indicated the sensor had begun 

fluctuating. This pressure was recorded. A plot was made of blade sensor 

pressure versus water cell pressure. A straight line was fit to the data 

points using a least square method, this resulted in a high correlation 

coefficient (1.00) indicating the sensor should be very precise. 



67 



Only two tests were performed with the new blade. One of the tests 
was done with the low consistency modeling clay confined by the K-Test 
mold. The previously established procedure for clay testing was followed 
in this test, except for taking a pressure reading with the sensor 
before the blade was removed from the specimen. 

The results of this test are summarized in Table 3. The initial 
mold a indicated in the table was defined as the lateral stress acting 
on the specimen before insertion of the blade. Final a_ mold is equal 
to the stress on the specimen at the soil-mold contact during the time 
the stress sensor was being actuated. Blade O is the lateral stress 
acting on the blade calculated from pulling force, and sensor G is the 
pressure measured by the sensor membrane. 



Table 3. Summary of test results using a blade with steel diaphragm 

stress sensor 

1 psi = 70.31 gm/sq cm 

Mold a (psi) 
Material Initial Final Blade O (psi) Sensor O (psi) 

11.06 
10.52 



Results of the modeling clay test indicate a significant difference 
between the calculated blade lateral stress and the sensor pressure 
reading. Since only a small increase in mold stress occurred due to 
insertion of the blade, it would seem unreasonable to expect a 10 fold 



Soft Clay 


4.57 


5.42 


40.74 


Sand 


4.27 


13.24 


18.67 



68 



increase in stress around the blade due to its insertion. One reason 
for the discrepancy between calculated blade O and sensor a is the 
assumption the friction angle and adhesion of soil on steel measured 
for the mold apply to the blade. 

A test was also performed on sand compacted to 40% relative density 
with the new blade. The same general procedure was followed in this 
test as was previously outlined in initial sand test. The results of 
this test also are indicated in Table 3, and are in closer agreement 
than in the clay test. As shown in Figure 16, the stress at the blade 
center is of the order of two to three times higher than at the edges , 
and this difference is accentuated by the K-Test mold. It should be 
mentioned that it was not anticipated that the agreement of data in 
Table 3 would be close, because of the high K-Test mold elastic constant 
compared to soil, the stress concentrations and discontinuities in the 
mold, and the indirect means for assuming the blade O from pulling force. 
Further Sensor Designs 

Preliminary tests with the steel diaphragm stress sensors indicated 
two rather serious problems: sealing of the diaphragm, and puncturing 
when testing in dense sands. Any hole in any diaphragm renders the 
device useless, and replacement of the stainless steel was inconvenient 
and required long setting times . 

The search for a material that would resist puncturing by sand 
grains ended with the choice of Teflon-TFE. The flourocarbon resin 



69 



film 0.010 in. in thickness was found to be tough enough and also 
provided a good seal between the pressure and exhaust chambers of the 
sensor. Figure 22 shows photographs of a sensor with and without the 
Teflon diaphragm. The diaphragm is held in place by a brass press- 
ring, and is supported from behind by a porous brass plate. The diaphragm 
and press-ring are both flush with the surrounding surface of the blade. 
Calibration 

Initial calibration tests of the stress sensor were performed in a 
cell filled with water, increments of pressure being applied to the 
water and thus to the external surface of the sensor. The stress at 
each pressure increment was determined by the sensor and the resulting 
data set was plotted. 

The water cell calibration method was time-consuming and difficult 
to perform but was felt satisfactory when stainless steel diaphragms were 
being used. However, with the ease of replacement of Teflon-covered 
sensors, a more convenient calibration method was considered necessary. 
For this purpose a small chamber that clamps to the blade is placed over 
the sensor, and pressure is applied to the chamber. The sensor stress 
corresponding to each chamber pressure is evaluated and plotted. A 
typical calibration plot for the sensors on the stepped vane is shown 
in Figure 23. All three plots have high correlation coefficients. 

The calibration plots are virtually unchanged when the Teflon 
diaphragms are replaced. A statistical analysis of variance of many 
calibration tests with different diaphragms indicated that the slopes 
and intercepts were essentially constant at a 95 percent confidence 



70 





Figure 22. Photographs of teflon diaphragm stress 
sensor 



71 



60 - 



£ 50 



o 



40 



w 

H 
CO 



P 30 



E- 
U 

< 



20 



10 



/ 



/ 



/ 



/ 



* 



/ 



/ 



/ 



/ 



y 



/ 



/ 



/ 



/ 



y 



/ 



f 



/ 



/ 



/ 



/ 



/ 



* 



/ 



/ 



/ 



/ 



/ 



1/8" sensor 



a =1.04a -1.09: r=1.00 
o s 



+ 3/16" sensor 



a =1.03a -0.73: r=1.00 
o s 



o 1/4" sensor 

a =1.02a -0.08: r=1.00 
o s 



combined: a =1.03a -0.63: r=1.00 
o s 



/ 



/ 



10 



20 



30 



40 



60 



SENSOR STRESS, a , psi 



Figure 23.- Calibration plots for the teflon diaphragm stress 
sensors of the stepped VSS device. 

1 psi = 70.31 gm/sq cm 



72 



interval. Only periodic calibration checks are thus required when 
diaphragms are replaced correctly. 

In calibration tests it was found that the rate at which pressure 
is applied to the sensor prior to leakage has an effect on the results, 
too fast a rate causing the diaphragm to lift at an erroneously low 
pressure. A very slow rate is therefore preferable. However, a rapid 
test also is desirable. Therefore a uniform rate of approximately one 
psi/sec was chosen as a convenient yet slow enough rate. Any rate can 
be employed as long as the calibration is determined using the same rate 
as in the test. 

Extensive tests using Teflon diaphragms have been favorable. Even 
in dense sand the diaphragms resist damage, and with the press-ring fit, 
damaged diaphragms can be replaced rapidly with a minimum of equipment. 
The only problem encountered has been in the field in very cold weather 
the brass press-rings loosen due to thermal contraction. This problem 
can be alleviated by use of rings with a thermal coefficient of expansion 
the same as that of the blades, or by using a flush-head screw to hold 
the ring in place, Fig. 24 shows three sensors mounted on a stepped blade. 
Peripheral Equipment 

The support system required to operate the stress sensor is shown 
in Fig. 25. Gas flows through a regulator and pressure gauge and goes to 
the sensor. A flowmeter in the return line then detects when the 
diaphragm lifts and vents into the exhaust chamber of the sensor. The 



73 






Figure 24. Photographs of blade with teflon diaphragm 
stress sensors. 



74 



PRESS-RING 



TEFLON 

DIAPHRAGM 



STRESS 
SENSOR 




INTAKE 
LINE 



PRESSURE 
GAUGE 




EXHAUST 
LINE 



REGULATOR 



PRESSURE 
CONSOLE 



( co 2 ) 

PRESSURE SOURCE 



ATMOSPHERE 




FLOWMETER 



Figure 25 • Schematic of stress sensor pressurizing system 



75 



flowmeter in turn vents to the atmosphere. The set-up is quite satisfactory 
although the flowmeter is very sensitive to any surges in operator application 
of pressure. 



76 



LABORATORY TEST RESULTS 

Disturbance in Layered Specimens 

One assumption basic to the proposed method of extrapolating to 
obtain the soil in situ stress is that a functional relationship exists 
between the thickness of a blade and the disturbance its insertion 
causes. To evaluate this assumption layered sand or modeling clay 
specimens were constructed, tested, then sliced open to allow visual 
observation of disturbance caused by blade insertion. 

All testing was done in accordance with previously outlined pro- 
cedures. At the end of testing the specimens were sliced open perpendi- 
cular to the width of the blade at the mold centroid. Photographs were 
taken of the exposed specimen cross section. Later, measurements were 
taken on the photos to define the limits of disturbance. For the purpose 
of this study, disturbance was defined as the total distance from each 
blade face to the point at which horizontal layers deviated from their 
original position. 

A total of eight layered sand specimens and six layered clay specimens 
were constructed to evaluate disturbance caused by blade insertion into 
sand confined by a K-Test mold. Six of the sand specimens were compacted 
to a relative density of 40%, while the other two were at 75% relative 
density. Testing was done on 40% relative density specimens with rough 
and smooth blades, 1/16, 1/8, and 1/4 in. in thickness. The two denser 
specimens were used in tests with 1/8 in. thick, both rough and smooth 
blades. All test blades had angle tapers of 60% and were 2 in. in width. 



77 



Loose sand . The results of the 40% relative density sand tests 
are shown in Figure 26. A straight line was fit to each set of data 
points using a least square method. The equation of the line fitting 
the smooth blade points was: 

Disturbance =3.08 (Blade Thickness in inches) + 0.0385 (10) 

with a correlation coefficient of 0.999. The equation for the rough 
finish blade test was: 

Disturbance =3.26 (Blade Thickness) + 0.274 (11) 

With a correlation of 0.997. These high correlation coefficients strongly 
suggest that disturbance caused by blade insertion varies linearly with 
its thickness. 

It is interesting to note the intercept values of both lines, which 
indicate disturbance from an infinitely thin blade. The smooth blade 
line had a very low intercept while the rough blade line had a high 
intercept with the ordinate. The only major difference between the set 
of tests was the finish on the blades. Therefore it seems logical to 
assume more grain interlocking occurred on the rough blade surfaces, 
causing a large constant disturbance to occur. This insinuates that a 
set of tests done with very smooth blades should have a best-fit line 
with an intercept that approaches zero, because its finish would mini- 
mize interlocking effects on disturbance. 

Dense sand . Disturbances in the dense sand tests were found to be 
larger than in the loose sand tests. The smooth blade test results showed 



78 



^ Rough Blade 



A 



Smooth Blade 



1.0 



0.8 



CO 

w 

X 

O 0.6 

H 



w 
u 



Eh 
CO 



0.4 



0.2 




I 



1/16 1/8 

BLADE THICKNESS, INCHES 



1/4 



Figure 26. Measured disturbance in 40% relative density sand 
specimens. 

1 in = 2.54 cm 



79 



a 75% increase in disturbance, whereas the rough blade tests resulted 
in only a 40% increase in disturbance. But the actual values of dis- 
turbance of each test were found to be very close, 0.90" and 0.98", 
respectively. These results indicate that a disturbance limit may be 
approached which cannot be exceeded regardless of the blade surface 
roughness. 

Low consistency clay . Tests using both smooth and rough blades 
were performed on the clay layer samples. Figure 27 is a plot of blade 
thickness versus disturbance. As can be seen, the data points tend to 
be in a straight line, the same as was observed in the layered sand 
test. 

The best-fit line to the smooth blade data was as follows: 

Disturbance =1.96 (Blade Thickness) + 0.055 (12) 

A correlation coefficient of 0.999 was calculated for this line. 
The equation of the line that best fit the rough blade data points was, 

Disturbance =2.06 (Blade Thickness) + 0.051 (13) 

with a calculated correlation coefficient of 0.985. 

The intercept of the smooth blade line was found to have a higher 
value than the rough blade intercept, which is a reversal of what was 
found in sand disturbance tests. However, the values differ by only 8% 
and with the correlation coefficient computed from the rough blade data 
is low, below the 0.10 significance level. 



80 



Also shown in Figure 27 are average smooth blade data points 
obtained by first slicing sections of the smooth blade layer specimens 
at approximately 1/4 in. increments beginning at the mold centroid, and 
measuring disturbance on each section. 

Resulting disturbance data are shown in Figures 28 and 29. These 
measurements were then averaged for each blade thickness and plotted 
in Figure 2 7. 

The line best fitting the average data points is: 

Disturbance =2.10 (Blade Thickness) + 0.016 (14) 

with a correlation coefficient of 0.999. This line comes closer to a 
zero intercept, indicating probable mold effects. 

Stress Concentration 

Linear elastic theory based on Mindlin's derivation (15) indicates 
an extremely large increase in stress will result when a blade is 
inserted into a semi- infinite mass. This also will reflect in an increase 
in total stress in the K-Test mold, which simulates an elastic continuum. 
The amount of this increase is simply predicted by assuming that blade 
insertion in the K-Test mold causes an increase in the overall stress 
related to volume of the blade. The test results reveal that these pre- 
dicted stresses did not develop to such high magnitudes, probably due 
to plastic behavior around the blade and soil densif ication occurring in 
the specimen. If the soil were to undergo zero volume change, calculations 



81 



0.5 _ 



0.4 — 



CO 

Pi 
W 

Eh 
W 
§ 

a 



H 0.3 — 



w 
u 

m 

ID 
CO 



Q 



0.1 




0.2 — 



1/16 1/8 

BLADE THICKNESS, INCH 



1/4 



Figure 27. Measured disturbance in the low consistency 
clay specimens 

1 in - 2.54 cm 



82 



u 



w 
u 



D 
Eh 
CO 
H 
Q 



0.6 



0.5 



0.4 




5 0.30 
3 




DISTANCE FROM MOLD SLIT, INCH 

Figure 28. Plots of disturbance as a function of location 
in the low consistency clay. 
1 in = 2.54 cm 



83 



1 0. 15 .— 



w 
u 



H 

Q 



0.10 — 



1/16" BLADE 



DISTANCE FROM MOLD SLIT, INCH 



Figure 29. Plots of disturbance as a function of location 
in the low consistency clay. 

1 in = 2.54 cm 



84 



show that the mold stresses should have been approximately 5 to 6 times 
greater than actually measured for both the sand and for the clay. 
Typical data plots are presented in Fig. 30 and 31. 

The testing done on sand compacted to 75% relative density gave 
lateral stresses and stress ratios about three times as high as in the 
40% relative density sand tests. Since the gradation, shear strength, 
and friction of soil on steel values do not significantly differ between 
all tests, it is logical that the three-fold increase is due to differing 
soil compressibilities. 

Most testing was done with 2-inch wide blades. Tests were performed in 
40% relative density sand or soft clay with smooth one- inch wide blades. 
Results were approximately equal to those observed with a 2-inch wide 
smooth blade, indicating that the soil densif ication occurs mainly 
normal to the blade. 

In the clay tests the blade stresses appear to increase as the blade 
is extracted, and blade lateral stresses plot above mold stresses. Only 
initial (3 in. penetration depth) blade data points were used for extra- 
polation purposes. 
Stress vs . Blade Thickness 

A list of all tests performed with sand specimens is given in 
Table 4. The results of these tests are plotted in Figures 32 through 
36. Each figure has two plots, one with lateral stress as the ordinate 
and the other with stress ratio, X as the ordinate. All plots have 
blade thickness as the abscissa. It should be noted that 1/4 inch blade 



85 




■H 
CO 

ft 



CO 
CO 



EH 

CO 



w 

EH 



o 

H 
Eh 

CO 
CO 

EH 
CO 



10 




5 






1 1 1 ■ 1 1 1 



12 3 4 

DEPTH OF PENETRATION, INCH 



Figure 30. 



Test results from 1/8 in. thick smooth blade inserted 
into a 75% D sand specimen. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



86 



co 

CO 



Eh 
CO 



W 
Eh 
< 



200 




V 




• Mold 

* Blade 


20 — 


100 
















/*V 






10 — 








^r a 










-*" 











1 


i 


1 1 1 


1 



-H 

a. 



CO 
CO 



Eh 
CO 



W 
Eh 
< 



O 
H 
Eh 



CO 
CO 



CO 




2 — 



3 12 3 4 5 6 

DEPTH OF PENETRATION, INCH 

Figure 30. Test results from 1/8 in. thick smooth blade inserted 
into a low consistency clay specimen. 

1 in =2.54 cm , 1 psi = 70.31 gm/sq cm 



87 



400 




O 
A 


Mold 
Blade 






100 _ 


300 



























80 - 


LATERAL STRESS, 

O O 
O O O 












60 *_ 


___ 




O 

4 


o 

1 


1 


40 _ 
20 psi — 

1 



1/16 



1/8 



1/4 



15 












<< 10 










o 


H 










A 


STRESS 




o 


o 










t 


1 


1 


1 



1/16 



1/8 



1/4 



BLADE THICKNESS, INCHES 

Figure 32. Test results from 2 in. wide smooth blades inserted 
into 40% D r sand specimens. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



88 



CO 

CO 



H 
CO 



H 



300 



200 



100 




100 



_ 80 



_ 60 



- 40 



~ 20 



CO 



CO 
CO 

w 

H 
co 



i 

w 

H 



1/16 



1/8 



1/4 



o 

M 

H 



CO 
CO 



H 

CO 



10 



5 — 




1/16 1/8 

BLADE THICKNESS, INCHES 



Figure 33. Test results from 2 in. wide rough blades inserted 
into 40% D r sand specimens. 

1 in = 2.54 cm,, 1 psi = 70.31 gm/sq cm 



89 



300 



en 
en 
W 
Pi 
H 
tn 



200 



3 100 






O Mold 
& Blade 



1/16 



G 



1/8 



1/4 



— .80 



60 



40 



20 



2 

en 
en 

H 

cn 




BLADE THICKNESS, INCHES 



Figure 34. Test results from 1 in. wide smooth blades inserted 

into 40% D sand specimens 
r 



1 in 



2.54 cm., 1 psi = 70.31 gm/sq cm 



90 



CO 



en 

w 

H 
CO 



3 

w 



1000 


O Mold 
6 Blade 




o 


— 


500 













A 


e 




"— 




O 











1 


i 


1 1 





— 150 



— 100 



- 50 



•H 
CO 



CO 
CO 

Pi 

H 
CO 



w 

H 
<! 



1/16 



1/8 



1/4 



o 

M 
H 



CO 
CO 



H 
CO 




1/16 



1/8 



1/4 



BLADE THICKNESS, INCHES 



Figure 35. Test results from 2 in. wide smooth blades inserted 
into 75% D sand specimens. 

1 in =2.54 cm , 1 psi = 70.31 gm/sq cm 



91 



1500 


G Mold 










CO 


A Blade 






O 


— 


#i 












C/l 












g 1000 












p4 

H 












1 










— 


g 500 












H-5 




O 
1 


6 

1 1 


1 





200 



CO 



CO 

(*> 

w 
Pi 

H 

C/D 



00 g 

W 
H 



1/16 



1/8 



1/4 



2 

CO 

CO 

w 
p* 

H 
C/3 



40- 



20 











o 










A 




6 
1 


e 

_L 


1 


1 



1/16 



1/8 



1/4 



BLADE THICKNESS, INCHES 

Figure 36. Test results from 2 in. wide rough blades inserted 
into 75% D r sand specimens. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



92 



data points for the 75% density tests were extrapolated from a 1.5 inch 
depth of penetration to a 3.5 inch penetration depth so that all 
plotted values would represent lateral stresses and stress ratios at 
the same depth. 

Table 4. List of sand tests used in extrapolation procedure. 
1 in = 2 . 54 cm 



* 


Blade Width 




Blade 


Inch 


Blade Sur: 


A 


2 


smooth 


B 


2 


smooth 


C 


2 


smooth 


A 


2 


rough 


B 


2 


rough 


C 


2 


rough 


A 


1 


smooth 


B 


1 


smooth 


C 


1 


smooth 


A 


2 


smooth 


B 


2 


smooth 


C 


2 


smooth 


A 


2 


rough 


B 


2 


rough 


C 


2 


rough 



40 
40 
40 
40 
40 
40 
40 
40 
^ 40 
75 
75 
75 
75 
75 
75 

Blade thicknesses: A, 1/6 in.; B, 1/8 in.; C, 1/4 in. 



Visual inspection of these graphs indicates they cannot be 
described with a straight line fit. When this is done correlation 
coefficients are low, and the intercept of the line is usually a nega- 
tive number. As will be discussed later, exponential curves gave the 
best fit. 



93 



A listing of all tests performed in modeling clay is in Table 5, 
and the results of these tests are plotted in Figures 3 7 through 40. 



Table 5. List of clay tests used in extrapolation procedure. 
1 in =2.54 cm 



* 


Blad 


e Width 


Blade 


(inch) 


A 




2 


B 




2 


C 




2 


A 




2 


B 




2 


C 




2 


A 




1 


B 




1 


C 




1 


A 




2 


B 




2 


C 




2 



Clay 
Blade Surface Consistency 



smooth soft 

smooth soft 

smooth soft 

rough soft , 

rough soft 

rough soft 

smooth soft 

smooth soft 

smooth soft 

smooth stiff 

smooth stiff 

smooth stiff 



Blade thicknesses are the same as in Table 4 



Lateral stress plots of clay data were unpredictable. Since the final 
lateral stresses on the blade and mold were less than those encountered 
in sand, the initial lateral stress became more important, which would 
account for the apparently random behavior of these plots. The plots 
of stress ratio versus blade thickness are more consistent, and all 
plots except for the 1 in. wide blade seem to follow a general exponential 
function. 



94 






w 

Pi 

H 
CO 



i — i 

$ 
W 

H 





o 


Mold 










150 
100 


— & 


Blade 


Ck 




o 


" ' " 


50 




6 
O 

i 


o 

1 


1 


I 





40 



— 20 



1/16 



1/8 



1/4 



o 

M 

H 






H 



7.5 






\ 


O 


5.0 


— 




6 




2.5 




o 

JL 


O 
1 


1 1 



1/16 



1/8 



1/4 



BLADE THICKNESS, INCHES 

Figure 37. Test results from 2 in. wide smooth blades inserted 
into low consistency clay specimens. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



95 



cfl 



co 

w 
Pi 

H 
CO 



I— I 

w 

H 







O Mold 




A 


— 






a Blade 








300 










— 


200 




A 


A 


O 


— 


100 




O 
1 


o 

1 


1 1 


— 



— 100 



80 



_ 60 



_ 40 



20 



1/16 



1/8 



1/4 



en 

w 

H 
C/3 



10 


— 


a 


£ 


A 


5 








o 






o 
1 


O 

1 


1 1 



1/16 



1/8 



1/4 



BLADE THICKNESS, INCHES 

Figure 38. Test results from 2 in. wide rough blades inserted 
into low consistency clay specimens. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



96 



cO 
P-i 



co 
CO 

w 
p* 

H 
CO 



3 

w 

H 



150 



100 



50 — 





Mold 










6 Blade 




A 




— 




A 




— 


— ' 


A 




o 


- 




O 

1 


f 


1 1 





40 



20 



•H 
CO 



C/> 
CO 

W 

2 

H 
in 



H 

<3 



1/16 



1/8 



1/4 



CO 
CO 
W 
Pd 
H 
CO 



20 



10 




1/16 



1/8 



1/4 



BLADE THICKNESS, INCHES 

Figure 39. Test results from 1 in. smooth blades inserted into low 
consistency clay specimens. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



97 



cfl 

Ph 



CO 
co 
W 

H 

co 



S 

w 

H 
<! 











o 








O Mold 












A Blade 








150 
100 








A 


— 


50 





O 

1 


6 

[ 


1 1 





- 40 



- 20 



•H 
CO 



co 

H 



W 
H 

< 



1/16 



1/8 



1/4 



o 

H 
H 



CO 
CO 



H 

CO 



10 



— 






o 








A 




o 


o 






* 


1 


1 



1/16 1/8 

BLADE THICKNESS, INCHES 



1/4 



Figure 40, 



Test results from 2 in. wide smooth blades inserted 
into high consistency clay specimens. 

1 in = 2.54 cm , 1 psi = 70.31 gm/sq cm 



98 



Exponential Curve Data Fit 

Exponential curves were found to give the best fit to all stress 
vs. blade thickness data sets. A justification may be that as the 
thickness of the inserted blade increases, densif ication required to 
maintain a constant lateral stress must increase; since soils have a 
maximum attainable density, as compaction proceeds, more work will be 
needed for additional densif ication. 

A listing of best-fit curves for stress ratio versus blade thickness 
data from tests in sand is shown in Table 6, while Table 7 is a listing of 
the corresponding curves obtained from tests conducted in clay. All 
curves are in the general equation form: 

A- = ae bt (15) 

where t is the blade thi'.&.iess (inches) and A. is the same as previously 
defined. No curve fitting was done with the lateral stress data because 
of the variation in the initial stress. A value of a = 1 indicates 
extrapolation to zero thickness does give an accurate undisturbed stress 
measurement, and the value of b indicates sensitivity to thickness 
changes. 

It should be kept in mind while evaluating the strength of relation- 
ships, that correlation coefficients of 0.987 and 0.951 would indicate 
0.10 and 0.20 levels of significance, respectively. Most r values 
were above the 0.10 level of significance, indicating a strong relation- 
ship between blade thickness and mold stress ratios. 

Interesting results were obtained in the mold stress ratio curve 
fits. Actual values of "a" for sand did not vary significantly when 



99 



as 


H 


<t 


t^- 


o 


as 


o\ 


a* 


o> 


o> 


<r> 


as 


C7\ 


as 


cr> 



S2 
CO 



(0 
P 

m 

Q 

o 

•H 

p 

(0 

CO 
CO 
CD 

u 
p 

Cfl 

o 
p 

CO 

a) 
> 
p 

u 



-H 
P 

c s 

(U u 

o <* 

a Ln 

X • 



ro 



D 

xl 
iH 
O 

a 

p 



•H 



b 

cu 
x) 
co 

H 

PQ 



co 



CU 

a 
>> 

H 

cu 

"O 

co 

rH 

PQ 



CU 



o\ 
cu 
o 



u 

sO 

00 
CO 



4-t 



4J 



cu 



cu 



o 


r-i 


o 


CM 


CM 


II 


II 


ll 


II 


II 


r< 


r< 


^ 


r< 


r< 


rH 
<? 
Os 


CM 
CX> 


CM 

i-H 


m 

vO 

ON 


as 

ON 



r-- 


•U 


4-) 




i-H 


• 


iH 


i-H 


■U 


• 


r^. 


• 


• 


00 


O 


■H 


CM 


CM 


• 


iH 


cu 


rH 


iH 


r-- 


CU 


CM 


CU 


CU 


CU 


m 


i-H 


<r 


<r 


m 


iH 


O 


o 


o 


<r 


CO 


II 


ll 


II 


ll 


ll 


,-< 


r< 


r< 


<-< 


^< 


cu 




cu 


cu 




T) 


cu 


T3 


x) 


cu 


CO 


•n 


co 


cd 


T3 


rH 


cd 


■H 


rH 


co 


rQ 


43 


43 


43 


43 


42 




42 


42 




•U 


42 


■u 


■u 


4= 


o 


60 


o 


o 


60 


O 


d 


o 


o 


d 


e 


o 


e 


8 


o 


CO 


p 


CO 


01 


p 


C 


fi 


(2 


C 


c 


•H 


•H 


•H 


•H 


•H 



cu 

■a 

En 



P 
CU 

4-1 

cO 

a 



Q 
O 

t2 



m 

x) 

C2 



100 



>1 



U 



■P 

fO 

Q 

O 
■H 
-P 

(0 

en 
en 

CO 
Q) 
U 
-P 
CO 

o 

+J 

CO 
0) 

> 
u 



<0 

•H 
4-> 

c 

CU 
C 
O 

a 
x 
w 



■H S 

&-. CJ 

I 

co m 



4-1 

o 

>i 
u 
(d 



CN 



CO 

o 



en 
O 



O 
CU 

rH 
M 

O 



a) 
a 
>> 
H 

QJ 

CO 
rH 

PQ 



CTi vO CTi 

f^ 00 <Ti 

tT* CT\ CTi 



vO 
C7\ 



w 






4-J 


-t 


4J 


+j 


m 


• 


r^ 


CN 


• 


H 


• 


• 


o 


H 


00 


m 


rH 


CI) 


CU 


cu 


CU 


m 


m 


r^- 


r-% 


• 


• 


• 


m 


o 


o 


o 


O 


II 


n 


II 


II 


r< 


r< 


r< 


r< 


o 


r^ 


rH 


co 


00 


f"> 


VO 


vO 


CT> 


cr. 


m 


CTi 



4-1 


4-J 


4J 


CN 


<f 


CO 


rH 




• 


• 


• 


<r 


VO 


CN 


VO 


rH 


QJ 


CU 


CU 


cu 


r~- 


CN 


rH 


CN 


iH 


vO 


m 


d 


II 


II 


li 


ii 


<< 


r< 


r< 


^< 


0) 




CU 


cu 


T3 


CU 


t> 


TJ 


CO 


X) 


CO 


CO 


rH 


CO 


<H 


rH 


,0 




.O 


J3 


JS 




^ 


x: 


4J 


J2 


4-J 


4-1 


o 


bO 


O 


o 


o 


3 


O 


o 


6 


O 


e 


B 


CO 


U 


CO 


CO 


a 


c 


C 


a 


•H 


•H 


•H 


•H 



CN 



3 
CO 



ai 

rH 

(0 



^4 
CU 
4J 

CO 



CO 

rH 

U 



o 



CO 
rH 
U 

X) 

r4 
CO 



101 



compared at like sand densities, and for 40% D sands were close to one, 
which is very encouraging. The 75% D sand also had consistent, but 
higher a-values. The b-values for all sand tests were consistent, with an 
average of 10.2. 

The exponential curves found to best fit the clay test results are 
also indicated in Table 7. Correlation coefficients were found to be 
very high when mold stress ratios were related to blade thicknesses in 
both soft and hard clay. The a-values for these curves closely agreed, 
averaging 0.6 in magnitude, not far from the ideal value of 1.0. Similarly, 
b-values were consistent. 

Results of curve fits based on blade stress ratio for sand tests 
show somewhat lower correlation coefficients. This is probably due to 
the inaccuracies in determining the actual lateral stress on a blade. 
But even with a lower r value, general trends in "a" and "b" can be 
noted. A general trend in "a" and "b" coefficients for tests in clay is 
that a-values were low when smooth blades were used and high when rough 
blades were used, whereas "b" values showed the opposite trend. 

In general terms, it appears that an exponential function does a 
good job of defining the relationship between stress ratio and blade 
thickness for tests in both clay and sand. This type of function should 
theoretically work as well when used with measured lateral stress versus 
blade thickness data. Unfortunately due to limited testing data, it is 
impossible to establish an exact prediction equation for the values of 
either coefficient "a" or "b". 



10 2 



Stress Sensor Tests in K Test Mold 

Stress sensor development was pursued simultaneously with simple 
blade penetration tests in the K Test mold. The next step was to 
combine these efforts, and use blades with stress sensors. Since it 
was suspected that the assymetrical, overly stiff elastic confinement 
supplied by the K Test mold could have contributed error, in this 
series of tests the Teflon plate that previously was rigidly clamped 
on the top of the mold was held down with a spring, allowing some volume 
expansion of the soil upward. 

Blades 1/8 and 3/16 in. thick and 2 in. wide were pushed to depths 
of 3 to 3 1/2 in. in samples held in the K-Test mold. At the end of 
penetration, the soil stress acting on the blades was determined by the 
stress sensor incorporated in each. The blades were then pulled and the 
force necessary to pull was recorded, and a soil stress acting on the 
blades was then evaluated from the pulling force. 

Soils used in these tests included the two consistencies of molding 
clay previously used, plus three different grades of sand ranging in 
gravel content from to 26%, and a standard Ottawa sand. The clays were 
compacted as before, whereas all three sands were compacted to D = 40%. 
Relevant property data from K Tests and from grain-size analyses are 
presented in Table 8. 



103 



•H 
O 
U> 

P 
CO 
<D 

EH 

>i 

O 
-P 
ft} 
H 

o e 
a o 

<0 



co tr> 
cu 

-H r-l 

+j n 
Sh • 

0) o 

o 

U II 
A 
•H 

<-\ co 
u 

•H rH 
(0 
>i 

On 



00 

CD 
fH 

EH 




• CM vO 




P O 




CD 

p. 



CO 

0) 

01 

u 

(30 

cu 
Q 



•H 
CO 



p. 

CU 
4J 

S 



vO 
CN 



m 

CO 






OO 
CO 



00 

co 



vO 



o 



oo 
on 



c*g 



oo 



I I 



vO 
O 



vO 



CN 

on 



m 

vO 



o 
o 



o 
m 



o 
o 



oo 

vO 



o 

vO 



m 
vo 



m 
vo 



m 



ON 



m 

CO 



o 



XI 



Xt 



CN 

CO 



XI 



O 
vO 



O 
00 



m 



m 



-o 



m 



ON 
CO 



ON 
CO 



CT» 
CO 



m 

CO 



o 



cd 
U 



o 

CV3 



O 



cO 
O 



•H 
cn 



X) 
U 

cO 

X) Xl 

P C 
CO cO 

4J C/} 



I 

c/> 

a 



CN 
I 

CO 

a 



cd 

£ XI 

CO C 

■U cO 

•U C/D 

o 



m 

co 






^3 

CO CO 

C co 

o cu 

p o 

O iJ 

a 



CN 



o 



,0 iH 

■H iH 

CU «H 

ja H 
en 



co 

P. 

cu 

4J 
CU 

e 

CO 

u 

CO 



vO 
CN 



Ph 



CN 

CO 



HH* 



5-S 

m 



oo 



■u 

C 

CU 

p 
a 
o 
a 

cu 

P 

d 

P 
CO 
•H 
O 

e 

s 

i 



6-S 
CN 
CO 



h4 



m 



PI 



on 



O 
CN 



O 



w a 
o o 



cu 

CU 

p 

I CO CO 

O 4-) +J 

+j m m-i 

iH .O 43 



o 

CO 00 



CO 

cu 

•U 
O 

c 
cu 

X) 



p 

p- 

•H 

u 
a 
co 

43 

3 

C/3 



co 
o 



P 
•H 
CO 

c 

CU 
X) 

>> 
U 

XI 



cO 

a 



o 

o 



p 
•H 

CO 

C 
cu 
x) 

U 

XI 



104 



Results from these tests are given in Table 9. Blade stresses 
calculated from pulling forces also are included, and as can be seen, 
the ratio O ,/a of blade stress determined in this way compared to the 
readings from the stress sensors varies from 0.22 to 4.53, averaging 
from 0.56 for Ottawa sand to 1.48 for the other sands, and 1.90 for the 
hard clay to 3.86 for the soft clay. This large variability makes either 
the blade data or the sensor data suspect. The value and range of over- 
stress ratios A, and A calculated from the corresponding values are 
smaller for the sensor data, the mean values and standard deviations 

being A, = 5.0 + 5.0 and A =2.6+1.7. The sensor data therefore 
b — s — 

appear more reliable, but the problem of the K Test mold influences remains, 
The inconsistency of A with regard to blade thickness prevents fitting 
meaningful curves . 

In some of the sand tests, the values of A determined indirectly or 
measured directly by the stress sensors are less than one. The probable 
cause for this is a combination .of the shear behavior of sand and the 
close confines of the K-test mold. As the blades were being pushed into 
samples in K-test mold, observations were made at regular intervals of 
stress changes laterally and vertically in the mold as well as the force 
being applied to push the blade. In the tests on clays all of these 
values increased with penetration depth, but in some of the sand tests, 
all of these stresses and loads increased and then decreased at a penetra- 
tion depth of about 2.5 in. A decrease in the lateral mold stress would 
necessarily mean a decrease in the stresses acting on the blade. It seems 



105 



r-< 



^HHPINOIvOHtnH 
oo-3-r-ONincMOoo^d-cM 

OHOOHNCMvOHH 



co m rH o r^ oo oo 



o vo m cm o o o 



-tf -<r CO 
<r <r> r^ 



o o o 



r^ <j- co co 

MO OW 



O O rH O 



<< 



COOOCOvOrHCMCOOOm 



HNHHm<r<rooooM 



1^ lO (M H \0 H H 

m -cr vo rH m <r o 



rH H CM H O O iH 



oo r^ rH 
in co vo 



o o o 



CM r- iH vO 

rH vo o o\ 



rH O rH O 



to 



mocovoocsjr^inr-cri 
vovoinmmin<fco<r-<i- 



cm on m o co r-«. vr 
cm <f cti oo <r as oo 



vo r~- o 
r^- m cm 



00 rH CM vO 

vo co o as 



oooooooooo 



O^HHHHO 



O CM rH 



O O rH O 



COVDCOOCMrHt^Or^l^ 
OrHvOCOrHrHrHCMCMrH 



cr\ as r>> cm on co as 
cm vo r~- cm as m cm 



oo i^- CO 
-d- r-. rH 



vO rH <r oo 
as oo co cm 



vo <f m vo on i — vo vo in vo 



rH CM rH -vf CO 00 -vf 
CO CM CM 



CM VO <r 



co cm r-^ m 



o 



o^Hinvoooovono 
cMcovocMcMino^ocor^- 



ON00OrH00C0C0<J-rHCM 
rHrHrHrHrHiniHrH 



000 M» O-CfCO 
00 CM rH <f 00 CO rH 



m i^- rH co cm <f m 



oo co m 
cm vo <r 



CO CM CO 



o r^ oo oo 
oo <r vo <r 



m co mo 



W 



oo 
> 

cd 



HO~infOC\|h.N<fO>CM 

mcMooinvococMvooovo 



■o- cm m <r vo vo co 

NOCOOONOOCfi 



m cm vo 



co vo on 

O o\ h« 



co m >cr vo 
as as t~^ r~- 



•H 



to 



OOOOOOOOOO 

mininooomoom 



nNffir>no40^n<r 



o o o o o o o 
o o m o m m o 



U-l rH rH cm <r o CO 



o o o 
o o o 



co co in 



o o o o 
m o o o 



co co vo in 



H 

•H 



b 



vO O 
CM VO 



CO CM 



vO O vO <r m rH -<r 
CM vO CM OS CM CO ON 



CO CM CO rH m m rH 



I I I 



r-~ co vo 

ON ON OS 



m co r-» 



vo <r oo 
on as cm 



r> r-\ OS 



rH o- as co 

vo o> m ffi 



r- r~- ~* as 



•H 

co 



to 



03 

CO 
H 



O 

e^s 

o 

<r 

i 

x> 
fi 
cd 
oo 

CfJ 

cd 

4-J 
4-1 
O 



CO 


<* 


rH 


•si" 


j-\ 


as 


O^ 


co 


ON 


00 




ON 


CM 


r^ 


as 


CM 


o 


r^- 




<r 


CO 


r-» 




r~- 


o 


<t 00 


r-- 


as 


Os 


r*. 


00 


rH 


O^ 


r- 


vO 


o 




vO 


O 


CM 


o 


o 


m 


o 




vo 


rH 


vO 




rH 


CM 


vO vO 


vO 


CM 


r- 


vO 


m 


CO 


CM 

H 


•o 


CO 


m 


u 
o 

B-S 
O 

■vf 

1 

"0 

a 
cd 

00 


CO 


m 


-d- 


CM 

rH 


m 


o 
iH 


m 




m 


r^« 


m 




m 


m 


r-~ m 










H 


H 


m 


H 


H 


H 












H 


H 


H 




H 


H 


H 




H 


H 


H H 










m 


m 


o 


rH 


rH 


rH 


Ti 










rH 


rH 


rH 




rH 


rH 


rH 




m 


m 


rH rH 










o 


o 


o 


O 


O 


o 


u 










o 


O 


o 




O 


O 


O 




o 


o 


o o 










o 


o 


• 


• 


• 


• 


cd 










• 


• 


• 




• 


• 


• 




o 


o 


• • 










• 


• 


i 


1 


1 


1 


x) 










1 


1 


1 




1 


1 


1 




• 


• 


1 1 










I 


1 


\0 


VO 


VD 


vO 


a 




oo 


co 


co 


VO 


vo 


VO 


rH 


vO 


vO 


vO 


CM 


l 


1 


vO vO 


00 


CO 


co 


00 


oo 


CO 


rH 


H 


rH 


rH 


cd 

■U 


00 


rH 


rH 


rH 


rH 


rH 


rH 


1 

00 


rH 


rH 


rH 


1 

00 


00 


oo 


rH rH 


rH 


i — ! 


rH 


rH 


<-i 


rH 


CO 


CO 


CO 


co 


00 


rH,0 


^ 


rO 


CO 


CO 


CO 


a 


CO 


CO 


CO 


s 


rH 


rH 


co co 



106 



<< 



r Q 



^ 



60 
> 
cO 



pq 

•H 






CO 



■u 

CO 

cu 

H 



coin^fvomvocooocor^ 

oavr^ooi^oc^iNinin 

■H rH rH rH rH rH 



cN<-\o<fHNcoLnmn 

cnsfffiHOHiflO\NH 



CMCMCOCOCMvOCMCOCOCM 



•stNHOi^ON^dCTiOO 
•Cfffi^-COI^N^inOvO 



^oo<-cn<rcNro<f-<rcN 



OfMcn^OO>d-<rMN<f 

ioinm<fcMNincriONC^ 
oo^no<fnoMnLOO 



OOvONOOOCNN^flflO 

HoococococTivoHNn 



CO rH CTi >d" 0-\ vO CO O rH in 
rHiH rH CNiHiHrHiH 



|v CM rH rH CM CO r)<r CM N 
^0 00 N rs H<f CT> vD CO H 



oooooo oooo 
oooooo oooo 

lO <T rH CM CM vO CO M N Ifl 



<• c^ iv rv oov n is h cri 
on cm o\ cn nm \o cr\ m n 



HMAHvO-* commN 

iH rH rH 



co m crv <t cno ooi^or- 

vOvO <f fs vO<f CM m O rH 



in «Cf CM -CC <T ~3" 1TIN CON 



<fvDC?i<fvOO^Ovtai 
P1NP1NOCO«JO«f 



oo-o-mcMrv<i-a-ivo-<r 



co<t<rm<fcnoom 
r->vocriC3>cooo>*rv.m 



CNCMCMCM-vTCMr^CMCO 



mon^rONN<f^ 
ococoiN^inoicNM 



(OHHOHHCMMH 



NvOOON-vtON^LOOO 
•vf-r^<rOi— II^CNCTiVO 



tsHCMvocncorsOKt 

>* n n H co c co n co 



CMCMCOOOCOOvOOCO 

tnvONHOCMn<j^ 



mr>.NH<roocMcots 

rHrHr-HCMCMrHi-HrHCM 



ommomoor^rHO 

I — ^O^OOWVONNH) 



ooooooooo 
ooooooomo 

mcovOMomco<fH 



rNOOO\COHONCOO\ 
CMvOvOCNvOvOCMCMin 



i— lcMc\ir^voOr^cr\<r 



oorscnco<f(Ti<tvoco 

vOvOOrHmvOONCrilv 



invD\ONIC1^0CO<fN 













>1 




H 






H 




^ 


tr< H b* 


H 


H H H 


H 


co 




m 


H H 




m h 


H H 


crj 


ifluim 


LO 


rH rH rH 


rH 


rH 




o 


m m 




O rH 


rH rH 


rH 


O o o 


o 


o o o 


o 


CJ> 




o 


o o 




o o 


o o 


O 


o o o 


o 


1 1* 1 


1 


m 




1 


o o 




i r 


• • 

1 1 


■u 


00 | I | 


1 rH 


vO VO vO 


vO 


<4H 


00 


CO 


I I 


vQ 


vO VO 


VO vO 


'4-t 


^-.00 CO CO 


CO ~-~ 


rH rH rH 


rH 


•H 




-s*. 


00 00 


rH 


rH rH 


i-{ rH 


o 


1 — 1 ""V \ "^ 


-sCO 


"■V **-*. ^^. 


"^ 


4-) 


rH 


T-\ 


~^ "s. 


** — 


"■^ v*. 


^-*« "^ 


<Xi 


CO rH rH rH 


r-VQ 


CO CO CO 


CO 


C/> 


CO 


cO 


rH rH 


CO 


CO CO 


co co 



S 




H 


o 






u 




X) 


MH 




P 
CO 


X) 






CU 




CO 


■u 




CO 


CO 




CU 


rH 




p 


3 




^ 


CJ 




o 


rH 




•H 


CO 




4= 


o 




4-» 


. r, 




4-> 


CO 




o 


CO 




p 


CU 






u 


CO 


CO 


4-1 .O 

CO <1 


L -e- 


•H 


B 


p 


CU 


CU 


CO 


u 


X) 1 


+j 


CU 


CO 




XI 


rH CO 


rO 


4-» 


pQ pJ4 


< 


CU 


II 1 


1 


u 

CU 



CO 
CM CO 
CU 
U 
4-1 
CO 



•H 

CO 



O 

co to 

P '-- 

cu r 

co o 

II ll 



•H 
CO 



60 

CO CO ,o 



CD 
CO 
CU 
U 
4-> 
CO 



CO 
•H 
4-> 



II 



CO 
CO 

cu 

CO i-> 
CO 4-1 
CU CO 

u 

4-1 e 

CO O 

4-1 

P* 4-1 

o o 

4-1 ,0 



CO CO 
•H >H 



4-1 

•H 
G 



•H 
CO 



II O 



H PQ 
H 



•H 

CO rH r 
GOO 



<< 



P 
•iH 

P 
•H 

CO 
CO 
CU 

p 

A! 

o 

•H 

x: 

4-J 
CU 

,p 

4-1 

CO 

•H • 

6 

00 -H 

P X 

•H CO 

X) 

(U rH 
O CU 

cu cu 

5-1 4-1 
CL CO 

U CO 

CU CD 

XI CU 

E rH 

P P 

P -H 
CO 

•> 4-> 

Pi CD 
O 

rH CO 

VW CO 

(U £ 

H 

s 

ri 60 
O CO 

14-1 U 

CO Pi 

X) CO 

P -H 

CO X) 
4J 

CO CU 

H 4J 



H 
O 

53 



P 

•H 

in 

CM 

CO 

XI 
CU 

4= 
CO 

P 

a, 
cu 

XJ 

cO 



p 




•H 


• 




•rH 


O 


CO 


O 


a 


CO 


m 




a\ 


X) 


00 


CU 


• 


X 


vO 


CO 




3 


II 


P-. 






CO 


CO 


Pn 


<u 


W 


XI 





rO CO 



u 

4-) 

o 



o 

4J 

c 

rH O 

rH TH 

-< CD 

U 

CU 

- > 
•H O 

O 

m 

• -H 
CO CD 



X) 
CU 

CO 

3 



c 

•H 

CD 
P-, (U 

CO 
CU CO 
XI CU 
CO U 



jQ rQ CD 
CO CO -K 



10 7 



likely then that the effect of introduction of the blades in sand is 
made more severe by the small size of the K-test mold and the rigid 
confine of the mold base. 

The physical properties of modeling clay depend in particular on 
temperature. A Torvane, a pocket vane shear device, was used to evaluate 
the shear strength of the clay used in the tests. With a normal shear 
strength value of 8 psi at 23 °C, a one degree F change in temperature 
caused an 0.7 psi, or 9 percent change in shear strength. This 
effect would influence test results, since temperature changes were 
experienced during the testing program. 
Stress Sensor Tests in Test Box 

In order to examine the inferred limitations of use of the K Test 
mold, a set of tests was conducted with the vane stress sensors in a 
large box. In the initial tests the three sensor blades of varying 
thicknesses were alternately pushed horizontally into compacted soil in 
the box. In subsequent tests the stepped vane was used. A surcharge 
load was applied on the box to simulate conditions in a soil mass under 
overburden pressure. 

Soils . Monona loess soil, a uniform silt-sized material, and Shelby 
till soil (Table 8) were used in the box tests. The soils were mixed with 
water to the optimum moisture content and compacted to standard maximum 
density. Three layers with 220 blows of a modified Proctor hammer per 
layer were found to provide the necessary level of compaction. 



108 



Apparatus . The test box has interior dimensions as shown in 
Fig. 41. One end plate has a 3-in. diameter hole through which the 
sensors were pushed lengthwise into the box. A removable top plate is 
supported by a wide-flange beam to minimize bending of the plate. A 
thin layer of sand separated the soil from the plate. 

The entire assembly was placed on a Fairbanks scale with a frame 
such that a load can be applied to the scale platform. In this manner a 
surcharge load of 1000 lb was applied to the top plate of the box and 
transmitted to the loess soil, and 800 lb to the till. 

Test Procedure and Results . Following compaction of the soil, it 
was allowed to cure 24 hours before the surcharge load was applied, and 
another 24 hour period was then waited to allow the stress to equalize 

With the single blades a total of five sets of tests was performed 
throughout the length of the box. A hole was advanced between sets to 
enable fresh material to be tested. Each test set consisted of 
alternately pushing three thicknesses of blades in sequence from thinnest 
to thickest. Caution was taken to ensure that each blade was pushed in 
the void left by the preceding thinner blade. Because of its length, 
with the stepped blade only two depths were tested, and since a pilot hole 
was not drilled, difficulty was experienced pushing the blade deep enough 
for the third thickness. 

Table 10 summarizes the tests performed in the box. The theoretical 
stress at each location is calculated as the applied surcharge load 
distributed over the area of the box less the area of the soil removed 
by boring. Thus, the theoretical stresses on the single blades increase 



109 



TOP PLATE 



TESTING HOLE 




Figure 4L 



Sketch of test box 
1 in = 2 . 54 cm 



110 



X 




o 


£ 


m 


o 


4-> 


tf 


co 


co 


a) 


N. 


EH 





Sh -h 

O CO 

co ft 

c 

0) iH 

in 



co g 
CD o 

>H 
in m 

o 
II 

4-J 

-H 

in 



(0 



13 

CD 
4-> -H 

co co 

O 

Q-i " 

cO o 

4-1 CO 



T3 
CD 

u 
d 

CD 

CO 
CD 

S 



b 



13 -H 

CU CO 

•H Cu 
rH 

<d 



c ci 

O -H 
•H 

4J « 

co CD 

i-l o 

4-> C 

CD CO 

4J 

CD CO 

Pn -H 



CD 
XI 
CO 

r-\ 

PQ 



•H 
O 



cn 


<T> 


[^ 


CT> 


a\ 


CX> 


a> 


a* 


a\ 


CT\ 



LO 


VO 


vD 


r^ 


vD 


co 


m 


rH 


vO 




!H 


CN 


vO 


CN 


rH 


r~~ 


co 


co 


o 


. - 


• 


• 


• 


• 


• 


• 


• 


■ 


• 


CD 


O 


O 


O 


o 


O 


o 


o 


o 


o 


T3 
CO 



CN 



O 



00 
O 



CO 



CO 

o 



co 



CN 



CN 



o 






00 



00 



m 



r- oo 



o CN 

00 rH rH 



CO 00 CO 

vo oo <r 



ocNrH r-icN-d- o a\ as <r>x>tn cno 



moo m m 



co co <f 

rH CN 



m m in 
rH cn m 



co r- rH 

rH rH CN 



CN (^ rH 
rH CO CN 



r^» O 

r-\ CO 



C^ 00 CN 
CN O- rH 



00 O 
t-\ CN 









o 
m 



m 



CN 



m 



m 



co 



CO 



CN 



in 



CO 



o 

CN 



CN 



m 

CO 



in 

o 



m 
co 



m 

o 



<mu <dcq o <; pq u < ra u <! m u 



co 
c 
o 

c 
o 

s 



CO CO CO CO CO 

<; w o <; pq 



CO CO CO CO CD 

<C PQ O <3 PQ 






CD 
m 



V 



PQ 



00 

c 

•H 
CO 



5 

<r 

rH 
II 
O 

• r. 

CD 
T3 
CO 

rH 

CD 

00 

•H 
CD 



•H 



CD 
CO 



CO 



13 
CD 
P- 

a, 

II CD 

4-1 
PQ CO 

.«> CD 

CD X 
13 4-> 
CO 

rH C! 

X -H 

CD CO 
rH CD 
60 CO 



CO 
CD 

• o 

C iH 

•H X! 



00 



CD 

CO 

CD 
4-1 
rS 



60 

•H 
CD 



111 



with distance from the end since less soil in cross-sectional area is 
supporting the constant surcharge load. This still does not realistically 
describe the stresses occurring in soil along the sides of the hole, but 
may give a reasonable approximation of the stresses in the remaining 
undisturbed soil, which is of primary importance here. 

Exponential regressions were fit using equation (15) plus the 
definition of A.: 



X = f- = ae bt (15) 

o 



where a is the sensor stress and O is the initial stress. Hence 
s o 



a = aa e (15a) 

s o 



Extrapolation to t = gives 



a = aa 
o o 



It will be noted that in a regression of a vs. t, (acr ) and b_ are the 
regression coefficients. 

It can be seen in Table 10 that the values for a vary from 0.08 
to 1.3 for the loess soil, averaging 0.85, but with most values falling 
in the range 1.2 to 1.3. In the stiffer till soil the ratio is much 
higher, 3.1 to 4.2. No consistent difference was found from pushing 
successive single blades compared to pushing the stepped blade. All 



112 



measurements with the 1/4 inch portion of the stepped blade were 
believed to be erroneous since they were lower than values obtained 
with the thinner blade sections, and were not used in the analysis. 

The value of 1.2 to 1.3 for single blade tests in loess in the test 
box compares to a value of about 1.7 for soft clay tests in the K Test 
mold, supporting the view that the lower restraint offered by the box 
did result in a value for a^ closer to the ideal value of 1.0. Still, 
the close proximity of the blades to the unyielding sides of the box 
probably results in a larger stress acting on the sensor than if the 
soil were behaving as a semi- infinite mass. The adverse effects of 
confinement would be expected to be accentuated in tests on the glacial 
till soil, because of its stiffness and included pebbles, so we tend to 
disregard that data. If a_ should equal 1.0, equation (15a) becomes 



a = a e bt (15b) 

s o 



113 



FIELD IN SITU TESTS 

Methods and Interpretation 

Both single-blade and stepped blade tests were conducted in the 
field. Preliminary trials of a three-blade stepped vane are reported 
in the Appendix. 

Since the field in situ stress for the most part is unknown, or 
at best has been measured by another instrument whose accuracy is not 
known, it is tentatively assumed that the multiplier a in equation (15) 
equals 1.0, and equation (15b) applies. 

This assumption may not be entirely correct even in a semi-infinite 
soil mass because of dislocations of grains along the extrapolated "zero 
thickness" plane. Thus a should exceed 1.0, particularly in sands and 
coarse-grained soils, and some empirical correlation may have to be 
developed which considers soil particle size. Also, it is expected that 
any significant departure from the semi-infinite ideal condition such 
as proximity to a rigid structure or inclusion will boost a and 
give high readings. This possibility may be analyzed since an inclusion 
causing a high local value for a_ also should result in a high value for 
the stiffness parameter b. 

The method for field data evaluation therefore is as follows: 

1. If the measured stress (T s steadily progresses to higher values 
with larger blade thicknesses, an exponential regression of 0* s vs. blade 
thickness t is made in accord with equation (15b) . This gives the 



114 



following: 

(1) . Regression coefficient a . 

(2). Regression coefficient b_. 

(3) . Correlation coefficient x_. 
If r_ is below 0.98, the relationship may be invalid, and an alternative 
interpretation is made as follows : 

2. Two points may be selected to fit equation (15b). This gives: 
(1) . Regression coefficient O . 

(2). Regression coefficient b. 
If b_ appears unreasonably high or low for the material tested, the 
relationship may be invalid, and the third method for interpretation is 
made: 

3. A value for b_ is selected from adjacent test data and applied 
for individual points. In this case equation (15b) is rewritten 



a = o e bt (15c) 

o s 



This gives individual values for a but no other information. This method 

° o 

is perhaps most appropriate where a truly varies, and the data may be 

used to calculate a mean and range in G . 

° o 

Single-Blade Tests, Overconsolidated Loess 

The first field in situ stress tests were made with three thicknesses 
of blades pushed hydraulically into loess soil that had been overconsolidated 
by being overrun by a glacier. The test location is a 77-foot (23.5 meter), 



115 



approximately 2:1 roadcut extending down through Wisconsin till, the 
underlying Wisconsin loess, and into pre-Wisconsin till, Fig. 42. The 
cut is located about 2 miles southwest of Boone, Iowa. Loess density 
was approximately determined with an Eley Volumeter. The Wisconsin till 
dry density is known to be fairly uniform, about 1.86 (116 pcf ) . Loess 
data are shown in Fig. 4 3. 

The lower center of the 8-foot thick loess layer was bored 
horizontally for 2 to 2.3 meters, and the blades were pushed hydraulically 
into undisturbed soil at the bottom of the hole. The blades had circular 
Teflon-covered penumatic pressure cell which had been calibrated against 
air pressure. By pushing the blades with the flat sides oriented horizon- 
tally, vertical stress could be measured and compared to the calculated 
overburden stress; by pushing with the blades oriented vertically, 
horizontal stress could be calculated and used to find the lateral stress 

ratio K . 
o 

Results . Data from the Boone site are shown in Table 11. As can be 
seen, stresses from the thickest blade were lower than those from the 
intermediate blade, and therefore were omitted from the analysis. The 
thick blade was fabricated by attaching a removeable back-up plate to the 
intermediate thickness blade, and had less than the design tip-to-sensor 
distance. The two-point method outlined above therefore was used, giving 
a and b_, but no correlation coefficient. The values for b_ are in fairly 
good agreement, the higher value being horizontally, in the direction of 



116 




cd 




5 




o 


CO 


i— i 


u 




Q) 


r- 


+J 


01 


(11 


c 


R 


o 




o 


<T> 


PQ 


«tf 




n 


}-i 


ro 


rt 




at 


o 


C 






II 


<D 




+j 




•H 


+J 


UJ 


m 


•u 


■H 


en 




a) 




H 




# 




CM 




«tf 




<u 




u 




3 




00 





Pn 



117 



3 
oo 

00 
En 



5 

r- 

Pi 



O 

a; 

CO 



CO 






W 
Eh 
H 
CO 

Eh 

CO 

w 

Eh 

W 

O 
O 

m 















• — \ 


o 
o — 
















■H 


* 

Q 














O 


S3 














o\° CO— 


S3 

CO 














*t 
















w 
















N O 
















H U?— 






Eh 








CO 






H 








w 






CO 








J o 
















U "*— 
















H 
















Eh 
















PS 
















Cm O— 












m 






*" — 








CM 




hH 

u 












o ■ 






I* 
















Eh 
















W H U 
















J CO o 








in 








D S \ 








r- 








ffl W D> 








• 








Q 








H 








o 
















v£>— 
















O 
















LTT-^ 
















o\° 
















^ 
















Eh O 
















S3 "=3<— 


















W 


"-3-I- 
















Eh 


^T 


"T 






~T~ 








S3 








x 










o o 


















U on— 








1 


















J 


H X 


' 




« 


hH- 1 - 


iA.i 




-i_ 


(0 • 






w 


Cu 






4 


n u 






Eh 










3 . 






<: CD- 










■P s 






IS cm 










fd 
S3 






O 




































J 










a 










J 










J 






S3 




H 










H 






O 




Eh 










Eh 






H 




Eh 
















Eh 




PS u 










PS 






J Cm 




W fiC 










W Eh 






H H 




Pm Eh 






CO 




§ U 






O PS 




Pm S3 






CO 




o < 






CO U 




D O 






w 




J En 






CO 




1 U 






o 




1 S3 






w 




co 






l-H 




CO O 






Q 




CO 

w 
o 










CO U 

w 
o 






4-> 




1 


1 


I 


J 




I 1 


1 1 






O rH 


CN 


00 


<3< 




in >£> 


r- oo 




X CO 


















Eh H 


















Pm CD 


L 


n 


c 


) 


in 


> in 




w +j 




• 






# 


■ • 




a cu 


c 


D 


1 — 


1 


rH CN 


1 CN 




e 

















3 
o 

&l 

00 
H 

o 



o 

Oh 



CO 

u 

cu 
p 



e 

00 

o 

00 



M-l 



Figure 43. Soil data from the Boone, Iowa, test site. 

118 



the larger in situ stress. The vertical stress of 15.2 psi (104 KPa) is 
somewhat higher than the theoretical stress from elastic analysis, which 



Table 11. Field stress data, Boone site. 

1 ft = 0.3048 meters, 1 psi = 70.31 gm/sq cm 

Horizontal Vertical Horizontal o, . b, , 
Blade Penetration, f t . a , psi a , psi psi mm 

fa S 

15.2 0.293 



A(l/8 in.) 


6.5 


38.4 




B(3/16 in.) 


7.0 


61.1 




C(l/4 in.) 


9.0 


(50.4) 




A 


8.0 




54.3 


B 


7.5 




89.5 


C 


9.5 




(72.0) 



20.0 0.315 



K =||4=1.32 
o 15.2 



a assumed to equal 1.0. 



is between 12.4 and 13.0 psi based on an assumed till wet density of 
2.16 (135 pcf ) . If the vertical stress is 12.7 psi it has been over- 
estimated by a factor of 1.2, and a. in equation (15a) should be 1.2. 
This is in close agreement with the test box data on loess. 

The value for K does not depend on the coefficient a_ which affects 
both vertical and horizontal stress values. In an overconsolidated 

material K should exceed 1.0. The value of 1.3 therefore appears 
o 



119 



reasonable, and to the authors' knowledge represents the first field 
measurement of this coefficient. 
Single-Blade Tests, Underconsolidated Loess 

A loess testing site was selected in a 35-foot roadcut 2.2 miles 
east of Turin, in western Iowa on Iowa Highway 37. The cut is on the 
north side of the highway, entirely in Wisconsin loess. As at the 
Boone site, a hole was auger ed horizontally into the cut, Fig. 44. 
Blades were pushed by hydraulically jacking against a bearing plate 
anchored to the soil. The hole was advanced an average of 6 inches 
between successive tests with the individual blades. Soil data for 
soil from the testing depth may be briefly summarized as follows: 

~ T • • j tvi Particle Size Gradation 
Dry Liquid Plastic 

Density Limit , % Limit, % Sand Silt Clay 
1.60(100 pcf) 30.5 25.5 0.8% 87.4% 11.8% 

Results . Turin site stress data are presented in Table 12. The 
vertical stress data shows a consistently higher stress with thicker 
blades and an exponential regression gave r = 0.994, indicating accept- 
able data. Also the value for b_ is in the range found for the Boone 
site loess. 

The horizontal stress data do not show this consistent trend with 
blade thickness, and r_ is too low. The data therefore may be paired 
for analysis or, since b_ is known from other tests, analyzed as 
individual values. These methods for data treatment are shown in the 
lower part of Table 12. Of the three possible data pairs, AB, BC and 



120 







Figure 44. Test site - Turin, Iowa. N^, NW^, Se.13, 
R44W, T83N. 

1 ft = .3048 meters 



121 



Table 12. Results of Blade Stress Sensor Tests, Turin Site. 
1 ft = 0.3048 meters, 1 psi =70.31 gm/sq cm 



Blade 



Horizontal 
Penetration, ft, 



Vertical Horizontal 



a , psi 
s r 



a , psi 
s r 



psi 



b, 
mm 



-1 



A 
B 
C 

A 
B 
C 



8.0 


7.4 




8.5 


11.3 




9.5 


19.7 




6.5 




8.5 


9.0 




7.7 


0.5 




16.6 



2.72 0.308 0.997 



(3.8) (0.211) (0.800) 



Treatment as Data Pairs 



A 
C 
B 
C 



8.5 
16.6 

7.7 
16.6 



(4.4) (0.211) 



(0.8) (0.484) 



Treatment as Data Points 



A 
B 
C 



8.5 




3.20 0.308 


7.7 




1.78 0.308 


16.6 




2.35 0.308 
2.44 + 0.71 




K 
o 


= 0.897 



Parentheses indicate rejected values. 



122 



AC, the first is rejected because it gives a negative value for b. 

The other pairs also give unlikly values for b_, so the third procedure 

is used where points are evaluated individually with b_ assumed to be 

0.308. This gives a mean horizontal stress of 2.4 psi and K = 0.90 

o 

The vertical stress from elastic theory should be 15.5 to 16.8 psi, 
almost 6 times higher than measured. This difference probably relates 
to either inelastic behavior of the material or, more likely, stress 
relief by erosion and undercutting at the ditch level. This also would 
explain the relatively high K , but it should be emphasized that no 

other K data are available from either field or laboratory determinations, 

o J 

since this is an eolian silt soil that is underconsolidated as a consequence 
of never having been saturated with water. 
Single-Blade Tests, Alluvium 

Logan-1 . A site designated as Logan-1 is located just southeast of 
Logan, Iowa. The tests were conducted in a vertical hole on the Boyer 
River Floodplain. 

Soil description and data are in Fig. 45. Alternating layers of 
silts and clays were encountered throughout, densities and moisture 
contents being determined on samples sealed at the site and sent to the 
laboratory. Since all blade tests were run in a vertical borehole, 
only horizontal stresses were obtained. The hole was advanced between 
test in order to run each subsequent test in undisturbed material. 



123 





O 






rH 
















O 
















00— 












Q 

3 




o\° 












< 

CO 




















w o- 






EH 










N v£> 






^ 










H 






H 










to 






CO 










O 
















W ^J*~ 
















1-H 
















U 
















H 
















EH 




>H 












< CN 




a 












o. 




u 












o ■ 






>H 
















Eh 
















^ M O 


CM 






m 


in 


o 




J CO O 


m 






rM 


LD 


>>r 




D S \ 


• 






• 


. 


• 




m w en 


rH 






-H 


rH 


H 




Q 
















o 
















V0- 




















O 




















LO- 




















s 


OP 


^ 


















en 
























r- 


- O 






















Eh 


Eh ^t*— 
















< 














< 
















4 


IS 


Eh 






1 -J 










T 




CN 


S O 


< 


I 












L 




^ 


O f**- 


1 












4 




Oi 


u 


►A 














T 




^ 


tf 


CL, 


-- 










_ _i_ 




en 


W O j 
















■H 


Eh cn— ' 




Eh 












U 


















CD 
CO 


o 




S 


II' 










r*f 




















S 




















,-, 


s 






3 




m 








J 


s 


JT» 


z 




u 




u 








U 


*—* 


w 


o 


















Eh 


en 


M 

Eh 


Q 


< 




>H 








!" 
< 






hi CU 


3 


J 










J 


CO 


•• 


H H 


CO 


u 




u 






u 




w 


O tf 
















>H 


Eh 


CO U 


>H 


>H 




>H 






>* 


H 


M 


CO 


EH 


EH 




Eh 






Eh 


>H 


CO 


w 


^ 


hi 




hi 






J 


^ 




Q 


H 


H 




H 






H 


Eh 




CO 


CO 




CO 






CO 


u 


10 




















w 




















Eh 






V* >- 






V" 


^ V 




MH 


1 


1 1 




("•/V i 


1 1 


1 




1 1* 


1* f 


rH 


33 




















1 


Eh CO 


CN 


^T <£> 




CO 


O CN 


<tf 


*X 


CD 


O CN 


S 


CU >H 










rH rH 


rH 


i — 


rH 


CN CN 


< 


W CI) 


O 




< 


i d 


» 




db c 


O 


(J 


Q -P 


• • 


. 


• 




• 


• 


o 


cu 


H CN 


n ^ 


V 




in u 


) r* 


vA 


g 















Figure 45. Soil Data, Logan-1 TEST SITE 



124 



Results . Results from the Logan-1 tests are shown in Table 13. 
In the first set measured stresses decreased with increasing blade 
thickness, but as shown in Fig. 45, the tests were conducted in two 
different layers. In the second set the C blade gave a low value, and 
a two-point fit gave O =1.9 psi. This cannot be correct since it is 
less than the pore water pressure. 

Logan-2 . The erratic nature of the results was tentatively 
attributed to the alluvial stratification and not conducting sequential 
tests at the same depth. Since the stepped blade was not ready for field 
use at the time, a second site, Logan-2, was selected where tests were 
conducted in three different boreholes but at identical depths. Logan-2 
also is on the Boyer River floodplain, 5.5 miles southwest of Logan . 
Soil data from the three borings was quite consistent; data from one 
of the borings are shown in Fig. 46. 

Results . Horizontal stress results from Logan-2 also are shown in 
Table 13. The data were more consistent than at Logan-1, but again the 
thick blade data were suspect. Two-point fit gave stress values as 
indicated in Table 13. The values for b_ vary widely, as might be 
expected in a layered deposit. The calculated K values range from 0.5 
to 1.1, the high value being at a depth of 5 ft. These values are not 
unlikely because of the expansive nature of the clay mineral, but there 
has been no independent check on these results. 
Stepped Blade Tests in FHWA Test Pit 

The first "field" tests using the stepped blade sensor were per- 
formed in a test pit at the Federal Highway Administration (FHWA) research 



125 





o 






















o — 






















H 




























Q 






O — 
CO 






CO 
















o\° 






















o_ 






















v vD 










EH 












W 










t-4" 












IS) 










H 












H 










CO 












CO O.J 






















<tf 






















w 






















J 






















u 

H 0__ 


































>H 
< 








Eh cm 
PS 




















<; 














J 








ft 














U 








O — 






















X 
W EH 
J H 
P CO 

PQ S 
W 
Q 






o 
in 

rH 


1.45 
1.27 


1.30 
1.27 












O 






















vD— 






















O 






















IT)~ 
















1 




« 


o\° 




















3 






















CO 


h, 




















r- 


En O 




















En 


S <*~ 




























H 


J 








- - 




















< 




*• 


EH 


**] 
































5 


2 




















< 


















<* 


o o 




















1 
















•^ 


U ci — 






































PS 


PS 


0- 

PU 






< 


i ; 


r 


— 


- 

41 


« 
















-J— 




CN 


Eh O 












• 










rH 


5 












4-1 

St' 










u 
























0) 
CO 


O 
i-f— 












D 


H' 








r*» 
























w 


















*"-» 






a 






Cfi 




z? 








u 






•a 






d) 




u 








' — 






Jf* 


3 




iH 




1 — ' 














w 


o 




,Q 












>H 






CO 


H 


,G 


.q 




X 








^ 








Eh 


4-> 


<L) 




k 














J PU 


■H 


ft 




i-3 






u 






• • 


H H 


3= 






u 












w 


o p; 














>H 






Eh 


CO cj 


J 






>H 






Eh 






H 


co 


H 






Eh 






J 






CO 


w 

Q 


o 

CO 






(J 

H 






H 
CO 






Eh 




PU 






CO 












CO 




o 


















W 




Eh 


















Eh 
CM 






















4-> 


~T 






1 


1 


i 




1 


1 






1 




1 


BJ 


r-\ 






CN 


CO 


^r 




-O 


vQ 






r- 




<! 


Eh W 
Pm Sh 




























u 


w <u 




in 




o 


m 




O 






o 


Q -U 




• 




• 


• 




• 






J 


E 




o 




H 


H 




CM 







Figure 46. Soil Data, Logan-2 TEST SITE 



126 



• s 

CN O 

c b 1 

re to 

rrj ro 



o 

r- 



•H 

to 

04 



CU -P 



■P 

CO 



CO 

CD <sT 

'd o 

rH • 

pq o 



o 

to 

4J 
rH 

CO 

CU 



■a 



ii 






« 





CU 






u 


•H 


CU 


53 


CO 


r-l 


CO 


fX 





CO 




Ph 


0) 


* 




u 


3 




Ph 






CO 




a 


CO 




0) 


cu 




*T3 


u 




r-l 


■u 




3 


CO 


•H 


,a 




CO 


M 


iH 


CU 


0) 


Ct) 




£ 


4-> 

o 

H 





O CO 

to Cu 



CO 
4-1 -H 



o t> 



J2 

4-) • 

CU 4-J 

CU 14-1 



CU 

-o 

ct) 
rH 
PQ 



C 
crj 

o 

1-1 



CO 
CU 

Cu 

o 

CO 

CU 

u 

4-1 

cct 
60 
cu 
53 



CO o 



o 

CO 



CO 






CO 



va- 



in cni 



o 



oo 



CO 

CO 



cr> 
o 



CO 

oo 



vO CM 



-* o o 
ro v£> m 



m 



ON 

cr. 



vo oo 



CM 

vO 



O 
CTi 



O 
CM 



CM 



o o o 

CO CO oo 



o o o 
m m m 



pq cj> 



pq o 



CM 

I 
S 

60 
O 

►J 



PQ U 



PQ O 






m m 
o o 



o 


T-\ 


-d- 


m 


00 


o-\ 


<r 


CM 


o 


vO 


CM 


o 


• 


• 


• 


• 


• 


• 


CM 


CM 


CO 


co 


m 


m 



o 
co 
o 



CO 

r-. 



co 



rH CO rH 

<r <t r^ 



o o o 
r-~ r~ r~~ 



PQ O 



co 
•H 
CO 

cd 

co 
co 

CU 

u 



CU 

> 

•H 
4-1 

o 

CU 

iw 

CM 

w 



127 



facilities in McLean, Virginia. The pit is approximately 10 feet 
square with concrete sides. Previous studies in the pit by the FHWA 
were involved with subgrade performance, so there was an asphaltic 
concrete surface course approximately 4 in. thick covering the soil. 
A vertical hole was bored into the underlying soil to a depth of 27 in. 
The stepped blade was pushed into the soil at the bottom of the hole in 
increments of 6, 7 and 9 in., so the sensor of each successive blade 
section was at the same point in the soil. Soil properties to the 
depth of testing are shown in Fig. 47. This was a very firm soil, 
having been compacted to 100% standard density and subjected to many 
cycles of loading tests. The moisture content was well below the 
plastic limit. 

Results . Results are in Table 14. The thickest section of the 
stepped blade gave a lower stress than the intermediate blade. A 
two-point fit to the other data gave b = 0.443 mm , which appears 
reasonable for a dense, highly constrained soil, and gave a lateral 
stress of 11.3 psi. Dividing by the calculated overburden pressure of 2.3 
psi gives K =4.9, indicative of a highly overconsolidated deposit. 
Stepped Blade Tests, Mitchellville Till Site 

Horizontal stress evaluations were performed with the stepped blade 

in conjunction with subsurface exploration for cooperative grain storage 

44 
bins in Mitchellville, Iowa, approximately 20 miles east of Des Moines. 

The site is located geologically within the end moraine of the Cary 

substage of Wisconsin-age glacial till. 



128 



O 

H 
Eh 
<C 
Eh 
CO 

CO 



w 
o 

l-H 



ft 

Eh 
CO 

w 

Eh 
ft 



o 




o 














M 


> 












Q_ 
dP °° 


3 












SIZE, 
60 




Q 

CO 










3 c ^- 














H 














EH 


Eh 












2 « 


i-h 




>H 
< 

' U 








o- 




>H 














EH 














WHO 




ro 


00 








J CO O 




r- 


r- 








D 53 \ 




• 


• 








m Hi? 




i-H 


<-{ 








Q 














O 














vi>- 














O 














0\° LO- 


>-*1 
J 














EH 
















S o 
















W <** 
















Eh 
















3 
















o 
















u o 
















ro- 
















ft 


l-H 














w 


PT 














Eh 














< O 














S CM- 
















4 


< 


■ 4 


4 






CD 














































ffi 














cj 














' 










s 




Q 










O 




S 










H 




sC 










En 




CO 










ft 














J H 




X 










H OS 




w 










o u 




>H 










CO CO 




<: 










w 




ij 










Q 




u 












1 

O 




J 


™ 


1 

CN 


1 


J 


m co 
















Eh M 




u 




c 


Lf 


c 


LT) 


ft (D 




CM 


in 


r» 


c 


csr 


W 4-> 




• 








• 


Q <U 




O 


c 


c 


1- 


r-1 


g 















Figure 47. Soil Data, FHWA Test Pit 



129 



Table 14. Results of Stepped Blade Tests at FHWA Test Pit and Mitchellville 

Site. 2 
1 ft = 30.71 cm , 1 psi = 70.31 gm/sq cm 



Blade Depth, 
Section ft. 



Horizontal 

a , psi 
s 



a o> 
psi 



mm 



Pressuremgter 
P Q > Psi 



FHWA 



A 


2.5 


46.0 


B 




92.9 


C 




(92.3) 


Mitchellville 




A 


16 


26.5 


B 


(Till) 


32.9 


C 




38.8 


A 


31.5 


27.5 


B 


(Loess) 


32.4 


C 




42.0 



11.3 



0.443 



18.3 0.120 0.997 



17.7 0.133 0.992 



19.3 



16.7 



Pressuremeter data courtesy of Michael Feist and Soil Testing 
Services of Iowa, Inc. 



The glacial till therefore should not be heavily overconsolidated . Test 
borings showed about 18 ft of fill and till-alluvial mixtures, overlying 
Wisconsin loess to a depth of 45 ft, overlying a paleosol (gumbo til) 



130 



developed in pre-Wisconsin glacial till. The soil data are shown in 
Fig. 48. 

Two sets of stepped blade stress sensor tests were conducted to 
determine horizontal stresses, at depths of 16 and 31.5 feet. In 
addition, pressuremeter tests were run at the same depths to find 
corresponding p 's as a comparison. 

Results . The Mitchellville test sets appeared to give valid stress 

measurements for all blade sections, indicating that the thickest 

(1/4 in.) blade may perform satisfactorily in sufficiently compressible 

soils. Three-point regressions gave high correlation coefficients and 

reasonable values for b_ and horizontal stress, Table 14. Of particular 

interest is the close agreement to pressuremeter p values, within 1.0 

psi. 

The value for K is calculated on an effective stress basis, which 
o 

means subtracting pore pressure from both vertical and horizontal stress 
before finding the ratio. These calculations are as follows based on 
measured soil densities and depths below the water table: 



Depth, 


Overburden 
Total Stress 


Pore 
Pressure 


V 


V- 


K 


ft. 


, psi 

V 


u, psi 


psi 


psx 


o 


16(till) 


14.6 


1.7 


12.9 


16.6 


1.29 


31.5(loess 


) 27.9 


8.4 


19.5 


9.3 


0.48 



131 



o\ c 



w 

IS1 
M 
CO 

w 
u 

H 

& 

< 
ft 




Q 
CO 



Eh 

H 
CO 



A 



< 
u 



m 

CD 

Eh 



(N 
Pi 



O 

CO 



H 
3 



CO 



W 
Eh 
H 
CO 

Eh 
CO 

W 
En 

o 



w 

H 

> 

w 

S3 

u 

Eh 



, Eh 

y, h m 
J co o 
d a a. 

PQ W 

a 



ID 
O 



en 



o 



Eh 

H 

Eh 
2 
O 

o 

ft* 

w 

Eh 






o 
in- 



o 
-3H 



o_ 



II 



I 



■I 



2 
O 

H 

Eh 
ft 

^ S 

m ft 
o u 

CO CO 

w 

Q 



H 

fa 



CO 



CO 



U 



EH 

H 
CO 

Jh ^ 

W J 

SB 

u ^ 



Eh 

H 

CO ^-* 
CO 

>H CO 

w w 

>H O 



w 




u 




' 




X 


J 


3 


O 
CO 


o 


O 




w 


>H 


l-l 


H 


< 


u 


ft 


H 


^-^ 


CO 





^l -P 
ft 4H 
W 
Q 



O' 
CN 



7 



m 



Figure 48. Soil Data, Mitchellville site. 

1 ft = 0.3048 m, 1 pcf = 16.018 Kg/cu m 



132 



The K data suggest a rather surprising inference that the glacial 
till is preconsolidated whereas the underlying loess is not . The 
occurrence of soft, even under consolidated loess underneath till is a 
fairly common observation, and usually is attributed to the lack of 
free drainage for the loess while it was under ice pressure , whereas 
the glacial material contains random sand lenses. 
Stepped Blade Tests, Houston Clay 

Following the success at the Mitchellville site, stress measurements 
were made with the stepped blade in the highly expansive, over consolidated 
Houston clay on the University of Houston, Texas campus. The clay was 
not sampled, since the site was previously investigated in connection 
with on-going pile load test research. A soil description is given in 
Table 15. 

Table 15. Soil Data, University of Houston Test Site.- 
1 ft = 0.3048 meters 

Depth, ft. N, Blows/ft. Description 



0-9 5 Gray and tan 

Stiff to very stiff clay 
w/nodules below 2 meters 

9-12 10 Gray and tan 

Very stiff, very sandy clay 

12 - 27 12-18 Red and light gray 

Very stiff slickensided clay 
with calcareous nodules 

27-45 7-35 Light gray and tan 

Very stiff sandy clay with 
sand pockets 

Note: Groundwater at 7.5 ft. 

Data courtesy Prof. M.W. O'Neill and Fugro-Gulf , Inc. 



133 



Tests were conducted in duplicate at nominal 5-ft depth intervals 
in two different borings, giving 4 test sequences at each depth. A 
truck-mounted drill rig was used to wash-bore down to the testing depth 
and to push the blade hydraulically for three stress measurements at 
each depth. The two borings, each to 50 feet, required about 6 hours 
testing time for two technicians and a driller. 

Results . Stress data from the Houston tests are given in Table 
16. In none of the tests did the C (1/4 in.) section gives reasonable 
data, so all values of O and b are from two-point determinations. Where 
by inspection the b_ values were negative, a 's were not calculated, and 
where calculated b_ values appear too low or too high, data are put in 
parentheses . 

An aid to determination of whether b_ is reasonable or not is a plot 
of b_ versus depth, Fig. 49. Of interest is that the two borings show 
different trends, even though they are only a few yards apart in the 
same clay. Deleted values are usually high, probably reflecting 
proximity to nodular inclusions noted in Table 15. Stress data still 
may be salvaged by use of the one-point method with b_ values obtained 
from adjacent depths or read from Fig. 49. 

Averaged or "normalized" b_ data also may be used to back-calculate 
in situ stresses from each data point. This procedure assumes that the 
horizontal stress is more variable than the soil stiffness parameter b_ 
after deletion of extraneous b_ values; therefore b_ data are smoothed or 
averaged. This procedure also is reasonable when we see that variability 



134 



b, mm -1 



0. 0.2 



0.4 



T 



0.6/0 0.2 



Boring 1 



10 



b = 0.110 



15 



20 



+J 

*» 25 

K 
Eh 
ft 
H 
Q 



30 



35 



40 



45 



50 



0.4 0.6 



= 0.00892 D 
- 0.093 

r = 0.934 



Boring 2 



b = 0.01730 
+ 0.053 
r = 0.959 



- 0.0213D 
+ 0.657 



r = 0.984 



b = 0.964D 
- 0.371 

r = 0.982 



I? 



Figure 49. Houston clay stiffness factor "b" versus 
depth. 

1 ft = .3048 meters 



135 



e 

u 

>1 . 

(0 O 1 
<-\ in 

i 

C tr> 
O 

P r-\ 

to ro 



3 


•x 



w 

P -H 
CO 05 

Eh 

rH 
QJ 

TS - 
TJ . 

-i £ 

m 
co 

>3 ^ 

CL) O 

ft ro 

Q, • 

(U O 
4J 

CO II 



o 

CO 

-P 



CO 
(D 

Pi 



VD 



3 
Eh 



00 
C 
•H 
P. 
O 
PQ 



00 
C 

•H 
P 
O 

PQ 



•> vH 
O CO 



U 



•H 
CO 

P-. 



pq 



o co 

b a- 



CO 



u 



PQ 



P 

Oh 

QJ I 

Q 



co on 
vo <r 



co CM 
vO o> 



cr. 

CM 



co 

CO CTi 



vo r^ 
co co 



O CM 
rH CM 



CO <f 

rH o> 

rH O 

O O 



CO en 



co r^ 

rH CM 



t-H VO 



o r-. 

CM CO 



CM CO 
vO iH 
vO CM 



cm r-~ 

m o 
co co 



r^- m 
O cm 

CM CM 



CM 

O 

CM 



CM 

m 

CM 



m 

CO 



I I 



oo oo oo oo 






CM o 
w CM 



m 

o 



CO CO 

o <f 

O CO 

m m 



in 
m co 



CO co 

CM rH 
rH CM 



rH 
O 



r^ co 



vo co 
<t cm 






00 CO 
I-- rH 



m <r 
vO cr> 



<r co 
vo r-~ 



<-i VO 

rH rH 

m <H- 

CM 00 



rH <f 
rH 00 



aj 
p, 
o 

rH 

CO 

VO 
OJ • 
> CO 

•H vO 

4-J 

oo 

<D 
S3 



CO CO 
O CM 



CM r-i 



vO vO 

<j- o 



00 



CO 



o 

CM 



I I 



I I 



00 
CT> 



00 



CO 

o 



CO 



CO 
CO 



00 

1^- 



CM CM 

in rH 



VO 
00 CM 



00 ^-\ 
CO CTi 



<H- CO 
O CM 



m 

•-n CO 

cu m 
a • 
o o 

rH ^ 
CO 

QJ 

•H- 

p 
oo m 

QJ • 
S3 CO 



m 

CM 



oo 



QJ 

p. 
o 

rH 
CO 

QJ 
> 
•H 
•P 

03 
00 
QJ 
S3 



O rH 
CM O 



O O 



CO O 

-H- r- 
m m 



CM 

CO 

/-n CXi 
QJ O 
P • 
O O 

rH 

CO 

QJ 

■5 
P 

oo co 

QJ • 
S3 ON 



co r^ 
co oo 
cm -H- 



CM vO 

m o 

CM CO 



CM CO 

m vo 

<d- CM 



O -H- 

V0 r-t 
CO CO 



<f 00 r- CM rHvO HO 
i— I "— ' COCO rHCM CMCM 



CTi CM 



VO CM 

m m 



<t m 
o <f 



vo <r 
o m 



co <r 

CM vO 



co 
cm <r 



-o- CM 

CM O 



CTi 

r-~ vo 



vO CM 

r-i CM 



r-« co 



o oo 
co r~- 



00 VO 
ON O 



rH r^ 

CO CO 



-J" VO 

CO 00 



00 CM 

<r vo 



vo <r 
vo m 



l l 



l l 



CO ^j- 

rH O 

• I -I 

<|- CT> 

CTi v — 



-H- CO 



coo r-^rH r-~ r^ o oo i^ oo in vo invo h oo r^-rH vocm 
cmco rH-H- coin cTiin co cr. co-* r^-vo -<i- vo cmco h n 



r-] CO 

vo m 
o CO 



oo oo oo oo oo 



en o ctnco mcTi corn r^vo 



CO rH 



CM 


co 




r^ 




CO 




CTi 




On 




00 




00 




CO 




00 


CO 


CO 




co 




CTi 




CT> 




00 




00 




CT> 




CM 




a> 


• 1 


• 


1 


• 


1 


• 


1 


• 


1 


• 


1 


• 


1 


■ 


1 


• 


1 


• 1 


co 


vO 




CTi 




CT> 




rH 




CM 




vO 




m 




vO 




m 


CM 






m 


/— "s 










^r 




r~- 




<r 




m 




-4" 


rH <r 


CO 


co 


^-\ 


CO 


VO 


CO 


CO 


CO 


rH 


CO 


co 


co 


vO 


CO 


00 


CO 


00 CO 


co r-~ 


rH 


CO 


CO 


CO 


r^ 


o 


m 


m 


CO 


m 


vO 


CO 


00 


rH 


CM 


CT\ 


rH <r 


-Cf CO 


o 


VO 


r-^ 


m 


vO 


CM 


rH 


CO 


m 


r-i 


m 


o 


00 


-d- 


CO 


CTi 


CO o 


CM <d- 


rH 


m 


00 


CO 


CT> 


CT. 


00 


CM 


<f 


00 


CM 


-H- 


CTi 


ON 


rH 


CO 


m oo 



cm m 



oo m 
<r <t- 



m 

CM 


CO 
CO 


m 

CM 


CO 
CO 


m 

CM 


CO 
CO 


m 

CM 


CO 
CO 


m 

CM 


CO 
CO 


m 

CM 


CO 
00 


m co 

CM 00 


m 

CM 


CO 
CO 


m 

CM 


co 

00 


m co 

CM CO 


m 


m 


O 


o 

rH 


m 


m 

t-H 


O 
CM 


o 

CM 


m 

CM 


m 

CM 


o 
co 


o 

CO 


m m 

CO CO 


o 


o 


m 

-d- 


m 


o o 
m m 



136 



or error in either of two stress readings affects b_, in effect 
doubling the likelihood of error in _b compared to error in individual 
stress readings. Table 17 shows stress data recalculated by the one- 
point method with b_ values from the regressions in Fig. 49. The 
stress data of Table 17 were examined for erratic values, which were 
omitted. In all, 18 of the 75 points were omitted, 6 because of 
erroneous or unknown b_ values. A plot of stress vs. depth is in Fig. 50. 

The plot in Fig. 50 shows a very strong increase in lateral stress 
at depths between 10 and 25 feet, then an abrupt drop in stress some- 
where between 25 and 30 feet depth. This range corresponds with 
the zone of slickensides noted in Table 15 from 12 to 15 feet. These 
features are indicative of shearing due to horizontal expansion pressures 
The allowable pressure should increase with depth because of the larger 
restraint to shear by overburden pressure acting as the minor principal 
stress. It also will be seen that the standard deviation is largest in 
the transitional stress zone, as might be expected. An increase in 
lateral pressure also is seen above the water table, probably as a 
result of desiccation. The cause for the apparent increase at 35 feet 
is not known, but might be speculated as being due to a relict feature 
caused by a former low position of the water table, evidenced by the 
mottled soil colors. Data are erratic at 50 ft., perhaps related to a 
soil change indicated at 47 ft. Furthermore the measured stress should 
not be less than the pore water pressure inferred from ground water 
elevations. 



137 



Horizontal Insitu Stress, psi 



10 



20 

~1~ 



30 



40 



50 



60 



70 



10 



15 



20 



-P 



K 25 

E-i 

W 
Q 



30 



35 



40 



45 



50 _ 




ZON 

or 

SLICK SlJSIDES 



Mean of Blade Data 
1 Standard Deviation 



Self-Boring Pressuremeter 



Calculated Pore Water Pressure 



Figure 50. Horizontal stress versus depth in Houston clay. Pressuremeter 
data coutesy of Fugro-Gulf, Inc. 

1 ft .3048 meters, 1 psi = 7031 gm/sq cm 



138 



Table 17. Houston Clay Stresses with Linearly Regressed b_ data 

1 ft = 0.3048 m , 1 psi = 70.31 qm/sq cm 







Boring 1 






Boring 2 






Depth, 
ft. 


b, 

-1 
mm 


A 

a , psi 
o 


B 

a , psi 
o 


b, 

-1 

mm 


A 

a , psi 
o 


B 

a , psi 
o r 


a 
o 

psi 


5.25 


0.110 


14.2 


14.4 


0.149 


17.5 


17.9 


19.2+4.6 


5.83 


0.110 


26.5 


25.9 


0.149 


18.8 


18.6 


10.25 
10.83 


0.110 
0.110 


12.0 
15.7 


33. 4* 


0.236 
0.236 


8.3 

19.5 


16.3 
19.0 


16.5+3.0 


15.25 
15.83 


0.110 
0.110 


* 
50.5 

25.1 


51.7 , 
20.9 


0.325 
0.325 


13.4 
20.6 


14.0 
20.0 


18.6+4.9 


20.25 


0.110 


56.4 


57.3 


0.221 


44.7 


43.7 


46.3+11.7 


20.83 


0.110 


55.3 


54.5 


0.221 


29.2 


29.4 


25.25 


0.155 


59.9 


* 

39.0 


0.115 


60.7 


16.8 


63.4+4.1 


25.83 


0.155 


65.1 


59.0 


0.115 


68.1 


67.8 


30.25 


0.200 


16.6 


17 ' 5 * 
31.4 


0.197 


19.0 


18.7 


19.4+2.8 


30.83 


0.200 


19.9 


0.197 


24.7 


n.d. 


35.25 


0.244 


38.8 


39.3 


0.290 


30.2 


28.4 


35.3+6.8 


35.83 


0.244 


39.8 


43.9 


0.290 


26.4 


n.d. 


40.25 


0.289 


19.3 


25.0 


0.383 


12.4* 


12.6* 


22.7+2.6 


40.83 


0.289 


24.8 


23.8 


0.383 


20.4 


n.d. 


45.25 
45.83 


0.333 
0.333 


23.2 
19.0 


24.2 
18.4 


(0.385?) 
(0.385?) 


(8.0?)* 
(9.3?) 


(15.2? 
n.d. 


; 21. 2+2. 9 


50.25 
50.84 


0.378 
0.378 


14.5 
13.7 


* 

8.8 

13.3 


(0.383?) 
(0.383?) 


(4.9?)* 
(6.8?) 


* 

(1.6) 

n.d. 


13.8+0.6 



Data omitted from calculations of mean and standard deviations 



139 



Table 18. Houston Clay Kq Data 

1 ft . = 0.3048 m , 1 psi = 70.31 gm/sq cm 





Pore Water 


Total 


Stress 


Effective 




Depth, 


Pressure 
psi 




psi 


Stress, 

a ' 

V 


psi 

V 




ft. 


a 

V 




a h 


K 
o 


5.5 





4.6 




19.2 


4.6 


19.2 


4.2 


10.5 


1.1 


8.8 




16.5 


7.7 


15.4 


2.0 


15.5 


3.5 


12.9 




18.6 


9.4 


15.1 


1.6 


20.5 


5.6 


17.1 




46.3 


11.5 


40.7 


3.5 


25.5 


7.8 


21.3 




63.4 


13.5 


55.6 


4.1 


30.5 


10.2 


25.4 




19.4 


15.2 


9.2 


0.61 


35.5 


12.1 


29.6 




35.3 


17.5 


23.2 


1.3 


40.5 


14.3 


33.8 




22.7 


19.5 


8.4 


0.43 


45.5 


16.5 


37.9 




21.2 


21.4 


4.7 


0.22 


50.5 


18.6 


42.1 




(13.8) 


23.5 


(-4.8) 


- 



Table 19. Soil Data, FHWA Fairbank Highway Research Station, McLean, 

Virginia Test site. 

1 ft = 0.3048 m , 1 psi = 70.31 gm/sq cm 



Depth, ft. 



Description 



N, blows/ft 



0-5 
5-22 
22 - 30 
30 - 42 



Clayey micaceous clay 
Micaceous silt 
Micaceous silt 
Micaceous silt 



8-18 
19 - 26 
>27 



Note: Groundwater at 18.5 feet 



140 



Also shown in Fig. 50 are P data, courtesy of Fugro-Gulf, Inc., 
obtained from tests with a self-boring pressuremeter . Except at 50 
feet depth, the data are closely comparable. The pressuremeter p Q is 
read only to the nearest 5 psi, and in all cases except at 50 ft the 
agreement is within this range. Due to the larger test depth intervals, 
the pressuremeter apparently missed the peak stress zone at 25 feet. 

The stress data of Fig. 50 also may be plotted in terms of K , or 
the ratio of horizontal to vertical effective stress. The latter was 
done by assuming a uniform soil dry bulk density of 1.92 (120 pcf ) . 
Calculations are presented in Table 18 and shown graphically in Fig. 51. 
The maximum value of Kq is 3.5 to 4.0, at depths of 20 to 25 feet. 

In summary, the stepped blade data give reasonable values for 
horizontal in situ stress, and are closely comparable to self-boring 
pressuremeter data. 
Stepped Blade Tests at Fairbank Highway Research Station, McLean, Virginia 

Tests were conducted in two adjacent boreholes in a micaceous, 
silty residual soil at the FHWA Fairbank Highway Research Station, McLean, 
Virginia. A drill rig was used, the hole being advanced by augering. Soil 
data are presented in Table 19. 

Stress measurements made with the stepped blade at Fairbank are 
presented in Table 20. These data are averages of two or three pressure 
gauge readings, often by different operators interested in learning the 
technique. No usable data were obtained from the thick (1/4 in.) blade 
section due to firmness of the soil, so data are not included. 



141 



1.0 



2.0 



3.0 



4.0 




Figure 51. K versus depth for Houston clay. 
1 ft = . 3048 meters 



142 



Table 20. Results of Stepped Blade Tests, FHWA Fairbank Highway Research 
Station, McLean, Virginia Test Site. 

1 ft = 0.3048 m , 1 psi = 70.31 gm/sq cm 







Bor 


ing 1 






Bo 


ring 2 






V pi 


si 


V 

psi 


b, 

-1 
mm 


a s , p 


si 


V 
psi 


b, 

-1 
mm 


Depth 
ft. 


A 


B 


A 


B 


4.25 










15.0 


30.8 


3.6 


0.453 


4.83 










15.0 


36.0 


2.6 


0.551 


5.25 


26.8 


39.8 


12.2 


0.249 










9.25 










28.0 


54.0 


7.5 


0.414 


10.25 


30.7 


52.0 


10.7 


0.332 


36.0 


53.5 


16.3 


0.250 


10.83 


32.0 


51.5 


12.4 


0.300 










14.25 










42.5 


77.5 


12.8 


0.378 


14.83 










36.0 


62.5 


11.9 


0.347 


15.25 


40.7 


59.0 


19.4 


0.234 










15.83 


42.3 


65.5 


17.6 


0.275 










19.25 










44.0 


47.5 


37.8 


(0.048) 


19.83 










50.8 


81.2 


19.9 


0.295 


20.25 


35.7 


68.0 


9.84 


0.406 










20.83 


31.7 


83.0 


4.62 


0.606 










24.25 










38.5 


83.5 


8.2 


0.488 


24.83 










83.2 


123.0 


38.0 


0.246 


29.25 










103.2 


126.0 


69.2 


0.126 


29.83 










83.0 


164.0 


21.3 


0.429 


30.25 


40.0 


n.d. 


- 


- 










30.83 


41.0 


43.0 


37.3 


0.030 











35.25 90.7 127.5 45.9 0.215 
35.83 97.3 122.5 61.4 0.145 



143 



A plot of b_ values from Table 20 versus depth shows a wide scatter 
with no consistent trend. As previously discussed, the variability of 
b_ values is greatly magnified by their determination from stress 
measurements. In order to reduce this variability, data from the 
same approximate depths were grouped as shown by horizontal lines in 
Table 20, and regressions performed on the grouped data, Table 21. This 
gave much more consistent values for b_ which show consistent trends, 
Fig. 52- Low correlation coefficients do not necessarily indicate 
erroneous b_ values because of the data pairing, a high A blade stress 
normally with a high B blade stress, etc. The low r_ in the range 
29.25 - 30.83 feet is contributed to by Boring 1 data in a "soft zone," 
but omitting that data did not appreciably alter the b_ value. 

Stresses recalculated with b_ values from Fig. 5 2 are presented in 
Table 22 and plotted in Fig. 5 3, the + values and plotted ranges 
indicating standard deviations, Fig. 53 shows a uniform tendency for 
horizontal stress to increase with depth, as might be anticipated in 
a residual soil due to volume expansion on weathering although other 
factors, such as tectonic compressive stresses in the underlying rock, 
may have an influence. Two exceptions are noted — an apparent dis- 
continuity in the vicinity of the ground water table, and a low stress 
zone at 30.5 feet depth. The latter was described in the field notes as 
a "soft and soupy," possibly due to water concentration along a fracture 
or fault. 



144 



b, mm 



-1 



■p 



EH 

w 

Q 








0.1 


0.2 


0.3 


0.4 0.5 0.6 


0. 







1 


1 


1 


1 1 1 




5 


— 








/ b = 0.511 - 0.0200 D 




10 


- 












15 


- 








b = 0.224 + 0.00579 D 




20 


- 












25 














30 








' b = 


= 0.788 - 0.017 D 




35 















Figure 52. Fairbank Highway Research Station, McLean, Virginia 
site "b" values from grouped data. 
1 ft = 0.3048 meters 



145 



Horizontal Stress, psi 






ft 
w 

Q 








10 20 30 40 50 


60 







1 1 1 1 1 


1 


5 


- \ 


\ 3 

\ V, © Mean 




10 




VS. 6 
\K3d .j 1 Standard Deviation 

\ \ 4 Number of Test Points 




15 


— 


V\ 8 




20 


w. t. 


(CX-jc 6 




25 




\|- (J^l 2 




30 




\—i 1 X. N. 

"soft, soupy zone" ^s. Nv 




35 




W — Pore water pressure \^ \. 
1 \^|_0-^4 





Figure 53. Horizontal stress versus depth, Fairbank site. 
1 ft = 0.3048 meters, 1 psi = 70.31 gm/sq cm 



146 



Table 21. Grouped Data Calculation of b Values for Fairbank Site. 
1 ft = 0.3048 m , 1 psi = 70.31 gm/sq cm 



Depth 
Range , : 


Et. 


V 

psi 


b, 

-1 

mm 


n 


r 


4.25 - 


5.25 


4.8 


0.418 


6 


0.85 


9.25 - 


10.83 


11.3 


0.324 


8 


0.97 


14.25 - 


15.83 


15.1 


0.309 


8 


0.94 


19.25 - 


20.83 


13.6 


0.339 


8 


0.80 


24.25 - 


24.83 


17.7 


0.367 


4 


0.69 


29.25 - 


30.83 


24.8 


0.284 


7 


0.41 


35.25 - 


35.83 


53.1 


0.180 


4 


0.98 



147 



Table 22. Fairbank Soil Stresses with Linearly Regressed b Data 
1 ft - 0.3048 m 





Calculated 
mm 




Horizontal 


Stress, psi 




Depth, 
ft. 


Boring 1 
A B 


Boring 
A 


2 
B Aver . 


n 


4.25 
4.83 


0.428 
0.416 




3.9 
4.0 


4.0 4.0+0.1 
(9.6) 


3 



5.25 


0.408 


7.3 


5.7 






6.5+1.1 


2 


9.25 


0.330 






9.8 


11.2 


10.5+1.0 


2 


10.25 
10,83 


0.310 
0.299 


11.5 

12.4 


11.9 
12.4 


13.5 


12.2 


12.3+0.67 


6 



14.25 


0.307 






14.83 


0.310 


13.5 


14.3 


15.25 


0.313 


15.1 


13.3 


15.83 


0.316 


15.5 


14.5 



16.0 18.0 

15 . 0+1 . 5 8 



19.25 0.336 15.1 (9.6) 

19.83 0.339 17.3 16.2 ib.z+i.i 



20.25 


0.341 


12.1 


13.4 








20.83 


0.345 


10.6 


16.1 






13.1+2.0 


24.25 


0.370 






11.9 


14.3 





24.83 


0.367 






25.9 


21.4 


23 .7+3.2 


2 


29.25 
29.83 


0.290 
0.281 






41.1 
34.0 


31.7 
43.0 


37.5+5.4 


4 


30.25 
30.83 


0.273 
0.264 


16.8 
17.7 


12.2 






15.6+3.0 


3 


35.25 
35.83 


0.188 
0.179 


50.5 
55.1 


52.1 
52.2 






52.5+1.9 


4 



148 



The data in Fig. 5 3 also can be expressed in terms of K based 
on calculated overburden stresses. A dry bulk density of 1.50 (94 pcf) 
was assumed, and uniform moisture contents of 20% and 30% above and 
below the water table, respectively. These calculations are shown in 
Table 23, and results are plotted versus depth in Fig. 54. The soil 
for the most part appears to be overconsolidated, perhaps from reduction 
of overburden pressure through erosion. This would not explain the low 
K just below the water table elevation. Other possible explanations 
previously mentioned include expansion on weathering and relict tectonic stresses 



149 



Table 23. Fairbank Soil K Q Data 

1 ft = 0.3048 m , 1 psi = 70.31 gm/sq cm. 









Total 




Effective 




Aver. Depth, 
ft. 


Pore Water 
Pressure, psi 


Stress, 

a 

V 


psi 

a h 


Stress, 

a ' 

V 


psi 

V 


K 
o 


4.75 







3.7 


4.0 


3.7 


4.0 


1.1 


5.25 







4.1 


6.5 


4.1 


6.5 


1.6 


9.25 







7.2 


10.5 


7.2 


10.5 


1.5 


10.5 







8.2 


12.3 


8.2 


12.3 


1.5 


15.0 







11.8 


15.0 


11.8 


15.0 


1.3 


19.5 


0.4 




15.3 


16.2 


14.9 


15.8 


1.1 


21.8 


1.4 




17.3 


13.1 


15.9 


11.7 


0.7 


24.8 


2.7 




19.8 


23.7 


17.1 


21.0 


1.2 


29.5 


4.8 




23.8 


37.5 


19.0 


32.7 


1.7 


30.5 


5.2 




24.7 


15.6 


19.5 


10.4 


0.5 


35.5 


7.4 




28.9 


52.5 


21.5 


45.1 


2.1 



150 



K, 



1.0 



2.0 



5 _ 



10 - 



15 _ 



01 

.£ 20 



u 

Q 



25 



30 



35 - 



- 




01 ^ 




1 










y 


- 








/ 


- 


W.1-. f 




1 


— 


"soft zone" 


? \ 


- 







Figure 54. Ko versus depth for Fairbank site, 
1 ft = 0.3048 meters 



151 



CONCLUSIONS 

Blade Design : 

1. A 60 degree apical wedge angle minimizes stress increases in soil 
caused by blade insertion. 

2. Smooth blades produce less disturbance than rough blades. 

3. The sensor should be located approximately 15 blade thicknesses 
from the end of the blade to minimize effects of the active shear zone 
developed around the blade. 

4. The Teflon-diaphragm pneumatic stress cell developed in this 
research is thin, precise, accurate, sensitive, and simple in operation. 
The fabrication costs for thin blades to carry several cells are high, 
and consideration should be given to use of an electrical strain-gaged 
diaphragm. 

Test Results : 

5. Soil stresses on the blade are an exponential function of blade 
thickness. Thus measuring stresses with two different blade thicknesses 
allows extrapolation to give the stress on a zero-thickness blade. 

6. In hard soils the thick (1/4 inch) blade consistently gave lower 
stresses than did the thinner blades. This is not in keeping with the 
exponential relationship, and indicates a change in soil behavior with 
the thicker blade. Future instruments therefore should omit the 1/4-inch 
blade if it is to be used in hard soils. 



152 



7. Tests with the stepped blade sensor are rapid and produce 

replicated data that is highly advantageous for statistical treatment. 

The following tentative procedure was developed for interpretation of 
test results: 

a. Each data set incorporating results from different blade 
thicknesses at a given position is fit to an exponential equation of the 
form 

bt 

a = a e 
s o 

where a is the blade stress, t the blade thickness, and O and b the 
s o — 

regression coefficients. The G also represents an initial estimate of 
zero-blade-thickness stress. 

b. The "b" slope data are examined for consistency, trends, and 
relation to soil type. Tests made to date indicate that b_ usually will 
be in the range 0.120 mm for soft soils to 0.480 mm for hard, dense, 
compact soils. If b_ values are highly variable or fall outside of 

this range or do not agree with observations relative to the soil, data 
within depth intervals should be pooled and exponentially regressed to 
obtain better estimates of b_. 

c. The b_ values obtained by regression are placed with individual 
stress readings to give a revised estimate of insitu stress: 



a - a e" bt 
o s 



153 



d. The in situ stresses may be graphed and treated statistically 
by depth increments, plotted versus depth, and/or used to calculate the 
coefficient of earth pressure, K . 
Comparative Data : 

8. Pressuremeter data from two of the test sites, one in glacial till 
and loess, and the other in expansive clay, indicate very close agreement 
with blade stress results. The blade is easier to use than the pressure- 
meter, particularly in the self-boring versions, and because of its 
statistical advantage is more precise. 

Future Designs : 

9. Based on results described in this report, a three-bladed 
stepped vane with 9 pressure cells has been successfully designed, built 
and tested in preliminary trials. Results with that device are appended 
to this report. 

10. For future blades it is recommended that a piezometer be installed 
to allow monitoring of pore water pressure, in order that effective stresses 
may be calculated instead of being inferred from the depth below the 
observed ground water table. 



154 



FIELD TESTING PROCEDURES 

(1) A hole is advanced to approximately one foot above the desired 
testing depth. 

(2) The blade is pushed hydraulically until the first section has 
fully penetrated the soil mass, a distance of 9 inches. The drill 
rod is then lifted to remove load from the blade by means of a 
slip- joint inserted between the drill rod and the blade. 

(3) The gas supply valve is opened to the first blade section. 

(4) The pressure regulator is turned slowly to provide a uniformly 
increasing pressure. 

(5) When significant return flow is indicated by the flow meter, the 
pressure reading is noted and the system depressurized. 

(6) Steps (2) through (5) are repeated for successive blade sections. 



155 



REFERENCES 

1. Alpan, I. "The Empirical Evaluation of the Coefficient Ko and 
Kor." Soil and Foundation , 7, No. 1 (1967), 31-40. 

2. American Society for Testing and Materials. 1969 Annual Book of 
ASTM Standards . Part II. Philadelphia, Pa. : American Society 
of Testing and Materials, 1969. 

3. Baguelin, F. , J. F. Jezequel, E. LeMee, and A. Lemehaute. "Expan- 
sion of Cylindrical Probes in Cohesive Soils." Journal of the 
Soil Mechanics and Foundation Engineering Division , 98, No. 11 
(1972), 1129-1142. 

4. Baguelin, F. , J. F. Jezequel, and A. Lemehaute. "Self-Raising 
Placement Method of Soil Characteristics Measurements." Proc. 
ASCE Speciality Conference on Subsurface Explorations for Under- 
ground Excavation and Heavy Construction , Henniker, N. H. , 1974, 
pp. 312-332. 

5. Baguelin, F. and J. F. Jezequel. "Further Insights on the Self- 
Boring Technique Developed in France." Proc. ASCE Speciality 
Conference on In Situ Measurement of Soil Properties , Raleigh, 
N. C, 2 (1975), 231-243. 

6. Baligh, Mohsen M. and Ronald F. Scott. "Quasi-Static Deep Pene- 
tration in Clays." Journal of the Geotechnical Engineering 
Division , 101, No. 1 (1975), 1119-1133. 

7. Bishop, A. W. "Test Requirements for Measuring the Coefficient 
of Earth Pressure at Rest." Proc. Brussels Conf. on Earth Pres- 
sure Problems , (1958), 2-14. 

8. Bjerrum, L., J. K. Nash, R. Kennard, and R. S. Gibron. "Hydraulic 
Fracturing in Field Permeability Testing." Geotechnique , 22, No- 

2 (1972), 319-332. 

9. Broms, B. B. and J. 0. Silberman. "Skin Friction Resistance to 
Piles in Cohesionless Soils." Sols-Soils , No. 10 (1964), 33-43. 

10. Brooker, Elmer W. and Herbert 0. Ireland. "Earth Pressures at 
Rest Related to Stress History." Canadian Geotechnical Journal , 
2, No. 1 (1965), 1-15. 

11. Carroll, Dorothy. Clay Minerals : A Guide to Their X-ray Identi - 
fication . Geological Society of America, Boulder Colo. , Special 
Paper 126, 1974. 



156 



12. Clark, J. I. and G. G. Meyerhof. "The Behavior of Piles Driven 
in Clay. II. Investigation of the Bearing Capacity Using Total 
and Effective Strength Parameters." Canadian Geotechnical Journal , 
10, No. 1 (1973), 86-100. 

13. Fox, Nathaniel S. "Plan, Program, and Feasibility Study: Deter- 
mination of the In Situ State of Stress in Soils." Proposal HCP- 
12, 1976. Marietta, Ga. : Soil Systems, Inc., 1977. 

14. Fox, Nathaniel S., Richard D. Barksdale, Richard L. Handy, and 
Gary D. Trott. "Measuring and Testing Techniques for Determina- 
tion of the In Situ States of Stress in Soils." Interim Report 
DOT-FH-11-9172. Marietta, Ga. : Soil Systems, Inc., 1977. 

15. Kezdi, Arpad. "Pile Foundations." In Foundation Engineering Hand - 
book , Chap. 19, H. P. Winterkorn and H. Y. Fang, Eds. New York: 
Van Nostrand Reinhold Company, 1975. 

16. Ladanyi, B. and W. J. Eden. "Use of the Deep Penetration Tests in 
Sensitive Clays." Proc. 7th International Conference on Soil 
Mechanics and Foundation Engineering , 1 (1969, 225-230. 

17. Lambe, T. William and Robert V. Whitman. Soil Mechanics . New 
York: John Wiley and Sons, Inc., 1969. 

18. Lutenegger, Alan James. The Iowa Continuous K-Test: A Laboratory 
Test for Measuring Lateral Stresses in Soils Induced by Vertical 
Applied Loads . Unpublished M.S. Thesis. Ames, Iowa: Iowa State 
University, Library, 1977. 

19. Massarsch, K. Rainer. "New Method for Measurement of Lateral 
Earth Pressure in Cohesive Soils." Canadian Geotechnical Journal , 
12, No. 1 (1975), 142-146. 

20. Massarsch, K. R. , R. D. Holtz, B. G. Holm, and A. Fredriksson. 
"Measurement of Horizontal In Situ Stresses." Proc. ASCE Specialty 
Conference on In Situ Measurement of Soil Properties . Raleigh, N.C., 
1 (1975), 266-286. 

21. Massarsch, K. Rainer and Bengt B. Broms . "Lateral Earth Pressure 
at Rest in Soft Clay." Journal of the Geotechnical Engineering 
Division , 102, No. 10 (1976), 1041-1047. 

22. Meyerhof, G. G. "The Ultimate Bearing Capacity of Foundations." 
Geo technique, 2 (1950), 301-331. 



157 



23. Meyerhof, G. G. "Compaction of Sands and Bearing Capacity of 
Piles." Journal of the Soil Mechanics and Foundations Division , 
85, No. 6 (1954), 1-30. 

24. Meyerhof, G. G. "The Ultimate Bearing Capacity of Wedge Shaped 
Foundations." Proc. 5th International Conference on Soil Mechanics 
and Foundation Engineering , 2 (1961), 105-109. 

25. Meyerhof, George Greffrey. "Bearing Capacity and Settlement of 

Pile Foundations." Journal of the Geotechnical Engineering Division , 
102, No. 3 (1976), 197-228. 

26. Robinsky, E. I. and C. F. Morrison. "Sand Displacement and Compac- 
tion Around Model Friction Piles." Canadian Geotechnical Journal , 
2, No. 1 (1964), 81-93. 

27. Robinsky, E. I., W. L. Sagar, and C. F. Morrison. "Effects of 
Shape and Volume on the Capacity of Model Piles in Sand." Canadian 
Geotechnical Journal , 1, No. 4 (1964), 189-204. 

28. Sanglerat, G. The Penetrometer and Soil Exploration . Amsterdam: 
Elsevier Publishing Company, 1972. 

29. Spangler, Merlin G. and Richard L. Handy. Soil Engineering . New 
York: Intext Educational Publishers, 1973. 

30. Taradash, Samuel. "Structural Supports." Tunnel Construction 
State of the Art and Research Needs . Mildred Clark, Ed. Transpor- 
tation Research Board, Special Report 171, 1977. 

31. Tavenas, F. A. "In Situ Measurement of Initial Stresses and Defor- 
mation Characteristics." Proc. ASCE Specialty Conference on In 
Situ Measurement of Soil Properties, Raleigh, N.C., 2 (1975), 263- 
271. 

32. Tavenas, F. A., G. Blanchette, S. Levoueil, M. Roy, and P. La 
Rochelle. "Difficulties in the In Situ Determination of Ko in 
Soft Sensitive Clays." Proc. ASCE Specialty Conference on In 
Situ Measurement of Soil Properties , Raleigh, N.C., (1975), 450- 
476. 

33. Terzaghi, Charles. "Old Earth Pressure Theories and New Test 
Results." Engineering News Record, 85, No. 14 (1920), 132-637. 

34. Terzaghi, Karl. Theoretical Soil Mechanics . New York: John 
Wiley and Sons, Inc., 1943. 



158 



35. Terzaghi, K. and R. B. Peck. Soil Mechanics in Engineering 
Practice , New York: John Wiley and Sons, Inc., 1967. 

36. Tschebotariof f , Gregory P. Foundations, Retaining and Earth 
Structures . New York: McGraw-Hill Book Company, 1975. 

37. Vesic, Aleksandar B. "Bearing Capacity of Deep Foundations in 
Sand." Highway Research Board, 20 (1963), 112-153. 

38. Vesic, Aleksandar S. "Ultimate Loads and Settlements of Deep 
Foundations in Sand." Bearing Capacity and Settlement of 
Foundations . A.S. Vesic, Ed. Durham, N.C.: Duke University 
Press, 1965. 

39. Vesic, Aleksandar Sedmark. "Expansion of Cavities in Infinite 
Soil Mass." Journal of the Soil Mechanics and Foundations 
Division , 98, No. 3 (1972), 265-290. 

40. Vesic, Aleksandar S. "Bearing Capacity of Shallow Foundations." 
In Foundation Engineering Handbook , Chap. 3, H.F. Winterkorn 
and H.Y. Fang, Eds. New York: Van Nostrand Reinhold Company, 
1975. 

41. Wissa, Anwar E. Z., R. Torrence Martins and John E. Garlanger. 
"The Piezometer Probe." Proc. ASCE Specialty Conference on In Situ 
Measurement of Soil Properties , Raleigh, N.C., 1 (1975), 536-545. 

42. Wroth, C. P. "In Situ Measurement of Initial Stresses and Defor- 
mation Characteristics." Pr oc. ASCE Specialty Conference on In 
Situ Measurement of Soil Properties , Raleigh, N.C., 2 (1975), 
181-230. 

43. Marchetti, Silvano. "In Situ Tests by Flat Dilatometer . " Journal 
of the Geotechnical Engineering Division , 106, No. GT3, (1980), 
299-321. 

44. Feist, Michael. "Pressuremeter vs. Laboratory Testing for Deep 
Foundation Settlement Analysis." Unpubli. M.S. Thesis, Iowa State 
Univ. Library, Ames, Iowa, 1980. 



159 



Appendix A 
Vane Stress Test Procedure 

Preliminary 

The final vane developed in this project consists of three blades, 
each blade with three thicknesses and cells, giving a total of 9 stress 
cells. The blades and cells each are numbered 1, 2, and 3. Gas lines 
and selector valves are numbered with two-digit numbers, the first digit 
designating the blade, and the second the cell, from thinnest to thickest. 
Thus line 11 comes from the thinnest section of the first blade; 12 from 
the middle thickness of the first blade, etc., up to 33, from the maximum 
thickness of the third blade. 

In addition each blade carries a manifolded exhaust line marked "E". 
Set-up Procedure 

1. Open the console lid and back off the pressure regulator so the 
handle is loose (Fig. A- 2). This is to prevent surge when the gas bottle 
is turned on. Failure to do this could result in rupture of sensors. 

2. Lift console plate by removing the middle thumb screw (Fig. A- 2, 
A-3) , and open the gas bottle valve (Fig. A-4). Replace console plate 
and screw. 

3. Connect marked lines to appropriate quickconnects in the console, 
11 to 11, etc., (Fig. A-5) . Be sure all valves are in the closed (down) 
position. 

4. Connect exhaust line manifold which connects to 3 lines, to the 
flow gauge desiccator chamber held in the lid of the case (Fig. A-6) . 

Open the flow gauge valve several turns. (This valve is kept closed during 
storage to preserve the silica gel desiccant.) 



160 



5. If not previously attached, connect numbered gas lines to 
corresponding numbered fittings on the blade (Fig. A-7). 

6. Drill a 6-inch diameter hole in soil, to just above the first 
desired testing depth. 

7. Connect vane to the slip connector (Fig. A-8) which in turn 
connects to A (or AW) drill rod with an adapter. Orient the vane in the 
hole and note the orientation on a sketch. Lower vane to rest on the 
bottom of the hole, being careful not to stretch or foul the gas lines. 

8. Mark off a 9-inch increment for hydraulic pushing of the vane by 
the drilling machine, and push the vane 9 inches. Then lift the drill rod 
about 1 inch, to remove rod weight from the vane. The slip connector is 
included between the vane and the drill string for this purpose. 
First-Position Readings 

9. All toggle valves in the console normally are closed (handles 
horizontal), and the pressure regulator handle is unscrewed to be loose. 
Open one valve (Fig. A-9) and slowly screw in the pressure regulator (Fig. 
A-10) while watching the gauge and the flowmeter ball. When the ball 
jumps (Fig. A- 11) read the pressure gauge (Fig. A-12) and immediately back 
off the regulator. Repeat this procedure one or two more times. Record 
all readings. Readings should agree to + 1 psi. An excessive flow "blow- 
by" usually damages the sensor diaphragm and it must be replaced; see below. 

10. Back off * the regulator until the gauge reads zero pressure, 
close the open valve and open the next valve (Fig. A-13) and repeat the 
test. Continue until all valves have been read. 



161 



Advance Readings 

11. Push the vane 7 inches for the second set of readings, lift off 
the rod weight and repeat the test for valves 11, 21, 31, and 12, 22, 32. 

12. Push the vane 9 inches for the third set of readings, and repeat 
the test for valves 12, 22, 32, and 13, 23, 33. Care should be taken not 

to overpush the device on the last test as this can cause a stress increase. 

13. Close all selector valves, back off the regulator, pull the vane, 
and clean it off. Auger bore to the next testing depth and repeat starting 
with step 6. 

Interpretations 

The above testing sequence gives pressure data at two depths 9 inches 
apart: The first is 6 inches below the bottom of the hole and is read on 
3 sensors on three blades. The second depth is 13 inches below the bottom 
of the hole and is tested on three blades but with only sensors 1 and 2. 
This procedure may be diagramed as follows: 



First push, read: 11, 21, 31. 

Second push, read: 11. 21, 31. 

12, 22, 32. <- 
Third push, read: 12, 22, 32. 

13, 23, 33. < 



<n 



< J 



Data combinations 
for 6-inch depth. 
'Data combinations 
for 13-inch depth. 



An abbreviated procedure may be used wherein only a single depth is 
tested : 



First push, read: 11, 21, 31.* 

Second push, read: 12, 22, 32.*- 

Third push, read: 13, 23, 33. 



Data combinations 
for 6-inch 
test depth. 



162 



Note that in each performance sequence only the first digit changes, i.e. 
the blade number changes but the thickness remains the same. 

Replacing the Sensor Diaphragms 

Sensor diaphragms are preformed Teflon discs. They may become cut 
or perforated through use, especially if sharp rocks or dense sharp sand 
are encountered. 

Improper diaphragm seating is indicated by consistently low readings. 

Cut or damaged diaphragms give no return and no reading. 

To replace a diaphragm: 

1. Insert the special punch into one of the three holes in the back 
of the blade opposite the damaged sensor, and tap lightly with a hammer 
(Fig. A-14). Move to the second and third holes and repeat. This forces 
out the diaphragm retainer ring. 

2. Clean off the perforated sensor face and wipe with light oil. 

3. Position the new diaphragm and retainer ring (Fig. A-15) and 
push into place with the C-clamp vise. Be sure the ring is fully seated 
and the diaphragm has not been cut by the ring edge. Full seating is 
important in that no part of the retainer ring should protrude above the 
face of the blade. 

It is a simple matter to replace the sensor diaphragm. When in doubt, 
replace the diaphragm! 



163 



Manufacture of Sensor Diaphragms 

1. Place Teflon sheet on the punch anvil (Fig. A-16) and set on 
guide ring (Fig. A-17). 

2. Drive special punch (Fig. A-18) to cut out disc (Fig. A-19) . 

3. Place disc on the forming anvil (Fig. A-20) . 

4. Place forming die on top (Fig. A-21), hammer sharply (Fig. A-22) 
and remove the formed disc (Fig. A-23) . 



164 







Fig. A-l. Assembly : Console pressure regulator is turned counterclock- 
wise so knob is loose, so no pressure will be applied during 
hook-ups . 




Fig. A-2. Remove middle screw to open console deck. 



165 




Fig. A-3 . Deck is lifted for access to C0„ gas cylinder. 




Fig. A-4 . Turn on gas. 



166 




Fig. A-5. Connect numbered lines to appropriately numbered quick- 
disconnects . 



if 




Fig. A-6. Connect return line manifold to desiccator 



167 




Fig. A-7 . If not previously done, connect numbered lines to appropriately 
numbered elbows on the vane. 




Fig. A-8. Vane is fastened to an special connector which allows drill rod 
weight to be lifted off the vane. Assembly is then attached to 
any conventional rotary exploration drilling machine. 



168 




Fig. A-9. Measurement : Toggle valve is opened to transmit gas pressure 
to a pressure cell. 



Fig. A™10. Pressure is slowly 
increased to the selected cell 
while the return flow is 
monitored on the flow gauge. 




169 




Fig. A-ll. A rising ball indicates return flow. 




Fig. A-12. The pressure gauge is immediately read and the reading recorded 



170 




Fig. A-13. Procedure is repeated for the next pressure cell 




Fig. A-14 . Diaphragm replacement : Remove press ring by carefully driving 
a small punch through three holes on the back of the blade. 



171 




Fig. A-15. Hold a new diaphragm and the press ring in place. The ring 
is pushed down with a special C-clamp (not shown) . 




Fig. A-16. Diaphragm manufacture : Teflon sheet is laid on the punch anvil 



172 




Fig. A-17 . Guide is placed on top, 




Fig. A-18 . Punch is driven. 



173 




Fig. A-19. Diaphragm disc falls through, 




Fig. A-20. Disc is placed on a special forming die, 



174 




Fig. A-21. Top of die is put in place. 




Fig. A-22. Die is hit gently with a hammer 



175 




Fig. A-23. Disc, now cupped at the edge, is removed, 



176 



Appendix B 
Data Interpretation — Three-Bladed Vane 

1. As discussed in the Report, in situ stresses are evaluated by 
means of exponential least-squares regression of the type 

bt 

a = O e 
o 

where 

-2 
a = measured blade stress, FL 

t = blade thickness , L 

b = a constant indicative of soil stiffness, L 

_2 
a = in situ stress for zero blade thickness, FL 
o ' 

The regressions are readily performed with a hand-held advanced 
programmable calculator or any of the more sophisticated computational 
devices. Tentative on-the-spot interpretations are possible, but the 
results should be carefully reviewed before final acceptance. 

2. The most important part of the interpretation is the evaluation 
of the constant b, which may be examined for reasonableness and 
consistency. The value of b varies from for soft, plastic silts to 
about 0.4 mm (10 in. ) for compact, dense soils, and even higher if 
the blade is pushed close to a rigid inclusion. A tentative procedure 
to evaluate b from three-blade data is as follows : 

(a) Perform exponential regressions on each blade data set. 



177 



(b) Examine the resulting correlation coefficients. Reasonable b 
data with high r values should be averaged. 

(c) Use the averaged b with individual cell pressure readings 
to obtain individual extrapolated zero-thickness pressures: 

-bt 

a .= ae 
o 

(d) Examine the resulting calculated stresses for trends. Data 
not showing directional or depth trends may be averaged to obtain mean 
stresses and standard deviations. 

3. This procedure is based on the assumption that the soil response 
to variable thickness blades, indicated by the stiffness parameter b 
(which is the slope of the exponential relationship), is more likely to 
be uniform than the individual cell pressure readings, which are readily 
influenced by nonhomogeneity in the soil. Enough data are generated at 
each testing depth that valid sets should be obtained to give b. Finally, 
the value of b should be examined to see if it is in the correct range 
for the soil tested. This gives a two-way check of data reliability: 
Is b reasonable; is O reasonable and/or possible? (For example, a a 
less than the pore water pressure is highly improbable.) 



178 



Example Calculations 



Data Sheet 

Location: Anita Iowa Elevator By: Handy, Lutenegger, Saye Date: 7-8-80 



Soil: Alluvium; stratified 



Hole Blade 
Time Depth Orientation 



4:00 pm 10.5' 



//IN 



Push 
Distance 

9" 

7" 



Pressure Readings, psi, 
Blade Blade No. 
Segment 1 2 3 



1 

r- 1 



9" 



4:25 pm 

(Data Sheet will repeat for subsequent tests) 



2- A 16,16 16,17 17,17 



20,20 20,20 20,20 
12,12 12,12 12,12 



12,11 15,15 16,16 
17,17 22,22 19,20 



Note that only two readings were taken at each cell position, since 
each data pair was in substantial agreement. The entire set required 25 
minutes, which included drilling to the new depth; thus to test a 50-foot 
hole at 5-foot depth increments should require about 5 hours plus set-up 
time, or essentially one working day. 

The first step in interpretation is to identify data sets, that is, 
test data that were obtained at essentially the same depth with sequential 
blade segments. These are shown as A and B above, and should be noted on 
the field notes. Set A has three tests at 9 in. below the bottom of the 



179 



hole; set B has two points at 9 + 7 = 16 in. = 1.33 ft. below the 
bottom of the hole. 

The second step is to draw a table of data for regression analyses, 
with average data readings reported for the data sets: 



Data x = thickness, y = average pressure, psi; Blade No 

Set Blade Segment in . (mm) 1 2 3 

A 1 1/8 (3.18) (20) (20) (20) 

2 5/32 (3.97) 16 16.5 17 

3 3/16 (4.76) 17 22 19.5 

B 1 1/8 (3.18) 12 12 12 

2 5/32 (3.97) (11.5) 15 16 



Inspection of the data is made to insure that sequential blade segments 
give consistently higher readings, and suspect data are set in parentheses 
and omitted from the analysis. 

The third step is to perform regression analyses with individual sets 

and blades or arbitrarily pooled data. Inspection indicates a consistent 

increase in stress from blades 1 through 3, but sets A and B are in fairly 

close agreement. Since statistical reliability is enhanced by pooling 

data, sets A and B in this case will be combined for each blade. Results 

are as follows : 

Blade 





1 


2 


3 


2 
r 


(0.88) 


0.97 


0.94 


r 


(0.94) 


0.99 


0.97 


n 


3 


4 


4 


a Q , psi 
b, mm 


(6.2) 
(0.22) 


3.5 
0.38 


4.7 
0.31 



180 



The values for b_ are inspected for reasonableness and consistency, 
the b_ values being indicative of a stiff soil. The low values for 
b_ and _r for blade 1 indicate the data should be reexamined. Since 2 
out of 5 blade 1 points have already been omitted, this test is aborted 
It will be seen that the high value for intercept <i for blade 1 relates 
to the relatively low slope b_, which in turn relates to the low value, 
17 psi, for segment 3. 

Results may be reported as follows: 



Blade: 1 

Orientation N 

Stress 

direction N90E 



Total Stress, 
psi 



n.d. 



2 
SW 

N30W 

3.5 



3 
SE 

N30E 

4.7 



Blades are numbered counterclockwise, and stresses are normal to 

blade directions. Thus blade 1, pointed N, measures stress in an EW 

direction, blade 2, pointed N120W, measures stress N30W, etc. With readings 

from three orientations, principal stress directions can be determined as 

for rosette strain gage data. 

Calculation of K . In order to calculate the coefficient of earth 
o^ 

pressure at rest, K , the vertical stress must be measured or calculated, 

o 

and vertical and lateral stresses converted to an effective stress basis. 
The water table at the above site was at a depth of 12.1 feet, or below 
the test depth, so total and effective stresses are assumed to be the same. 



181 



3 
If the soil unit weight is 120 lb/ft , 



a = 120 x (10.5 + 1) = 1380 psf = 9.6 psi, 

the average test depth for sets A and B being 1 foot below the bottom 

of the hole. Values for K = 0, '/a ' are: 

o h v 

N30E N30W 

K : 0.36 0.49 

o 



182 



Appendix C 
Test Report: 3 Bladed Vane 

1. Calibrations . Cells were calibrated after the field tests by 
use of a clamp-on chamber and air pressure. Cells operating correctly 
show a 1:1 calibration. 





Applied 
Pressure, 




Blade 




Av. Ratio of 
Meas: 










Cell 


psi 


1 


2 


3 


Applied 


1 


10 


10 


13 


10 


1.10 




20 


18.8 


22.3 


20.5 


1.03 




30 


31.2 


29.3 


30.7 


1.01 




40 


41.5 


40 


41.5 


1.02 


2 


10 


9 


11.2 


* 


1.01 




20 


17* 


21 


* 


1.05 




30 


23.5^ 


29 


* 


0.97 




40 


28.5^ 


35.7 


* 


0.89 


3 


10 


9.5 


11 


* 


1.02 




20 


19.3 


21 


* 


1.01 




30 


30 


31 


* 


1.02 




40 


40 


41.2 


* 


1.02 








Wtd. 


Average 


1.02 



Return line plugged; no calibration 
Leaky seal; data not included. 



Blade 1 cell 2 gave a low, curvilinear calibration characteristic of a 
leaky seal, and a correction graph was prepared for these readings. 



183 



Scatter in the other data is attributed to experimental error and 
non-uniform calibrating pressure. The average calibration constant 
is 1.02, or nominally 1.0. 

2. Blade Thickness . It should be emphasized that the three 
sections of the vane blades are in equal steps from 1/8 to 3/16 in. 
(3.175 to 4.762 mm) thick, rather than 1/8 to 1/4 in. (3.175 to 6.35 mm) 
for the single stepped bladed discussed in the body of this report. The 
reason for this was the frequent failure of the 1/4 in. (6.35 mm) section 
to give meaningful results in stiff soils. 

3. Soils. The 3-bladed stepped vane was tested in the field in 
three soils, an soft alluvial silt, a soft loess, and a hard, clayey 
paleosol (buried soil) underneath the loess. The first was selected for 
having a known anisotropic loading condition. The soils and sites are 
as follows : 



Location 
Anita, la. 



Site 



Menlo , la. 
Menlo , la. 



Soil 
Alluvial Silt 

Alluvial Silt 

Alluvial Silt 

Loess 

Paleosol 



Depths 



Comments 



10', 15', 20' Av. 50' from 
loaded grain 
bins. 

5, 10, 15, 20' 9.3' from edge 
of loaded grain 
bin 

5, 10, 15' Away from bins 

5, 10' Very soft 

17' Hard, expansive 
clay 



184 






4. Menlo paleosol data . Since the paleosol, a stiff, heavy, 
montmorillonitic clay, is closer in properties to previous soils tested, 
these results are presented first. Data are as follows, parentheses 
indicating corrected values for cell 1-2. Pressures are in psi. All 
data represent averages of 3 readings. 



Blade 



Depth Cell 



17' 1 

2 
3 

17 1/2' 1 

2 



39.5 


29.5 


34.0 


(41.5) 


35.0 


41.0 


43.0 


44.0 


55.0 


45.5 


30.0 


44.0 


43.0 


44.5 


51.5 



2 
Exponential regressions gave r and slopes b as follows : 







Blade 




Depth 


1 


2 


3 


17* 

17 1/2' 


1.36; 0.99 

— » — 


6.40; 0.99 
6.20; - 


7.70; 0.98 
5.04; - 



Dashes indicate negative slopes or insufficient data. (A minimum of 3 
points is needed for a correlation coefficient.) The slope data from 
blade 1 is dismissed as unrealistic for a dense clay, leaving average 



185 



b values as follows : 

Depth b, in. (mm ) 

17' 7.05 (0.278) 

17 1/2' 5.62 (0.221) 

Exponential extrapolations based on these slopes give the following 
pressures: 



Blade Orientation 



Depth 



ell 


1 NW-SE 


2 E-W 


3 NE-SW 




Aver. 


1 


16.4 


12.2 


14.1 




14.2 


2 


13.8 


11.6 


13.6 




13.0 


3 


11.5 


11.7 


14.7 




12.6 


er . 


13.9 


11.8 


14.1 


V 


= 13.3+1.7 



17' 



17 1/2' 1 22.5 14.9 21.8 19.7 



17.9 


18.5 


21.4 




19^3 


20.2 


16.7 


21.6 


°h 


= 19.5+2.9 



High values in each set are underlined and show no consistent trend. 
The data indicate that the lateral pressure is higher and more variable 
at 17 1/2' than at 17' depth, probably reflective of changes in clay 
content in the buried soil profile. 



186 



Approximate K values were calculated based on a water table at 
7.5' depth and y = 125 pcf: 







Stress, 


psi 




a » 

V 




Depth, ft. 


°h 


u 




V 


K o 


17 

17 1/2 


13.3 
19.5 


3.2 
3.2 




10.1 
16.3 


10.9 
10.9 


0.9 
1.5 



In this table, a represents the measured total stress, u the 

calculated neutral stress or pore water pressure, a ' the effective 

horizontal stress, and Q ' the estimated effective vertical stress. 

v 

These values of K , while >1, are low for an overconsolidated expansive 
clay, but may be relict from when the soil was buried by eolian loess 
silt, 20,000 years ago. 

This is the first and only known measurement of insitu horizontal 
stress in a buried expansive clay paleosol. 

5. Menlo loess data . The loess at Menlo in west-central Iowa is 
moderately clayey and very soft due to a moisture content well above 
the plastic limit, hole closure indicating a close proximity to the 
liquid limit. This is the first time such a soft soil has been tested 
by the vane stress device, and it may be anticipated that the stiffness 
parameter b will be very low and perhaps even zero. The data are as 
follows : 



187 





Cell 




Blade 




Depth, ft. 


1 


2 


3 


5 


1 


25.5 


24.5 


21 




2 


(28) 


20 


21 




3 


23 


23 


20 


5.5 


1 


25.5 


23.5 


22 




2 


(38) 


18.5 


25 


10 


1 


13.5 


11 


12 




2 


(13.5) 


11.5 


15 




3 


17 


16 


16 


10.5 


1 


16 


12.5 


12 




2 


(23) 


18 


18.5 



2 
Regressions of data sets gave the following values for b and (r ) 



Blade 



Depth, ft. 1 2 3 Aver, 

5 -1.65(0.27) -1.01(0.09) -0.78(0.75) J 
5.5 12.8 ( - ) -7.66( - ) 4.09 ( - )) 



1.0 + 6.9 



10 3.69(0.75) 6.00(0.84) 4.60(0.91) 4.76+1.16 
10.5 11.61( - ) 11.67 ( - ) 13.85 ( - ) 12.4 + 1.3 



188 



At 5 and 5.5 feet depth the b values often are negative and regression 
coefficients are very low, indicating that the soil is yielding and 
not building up stress as a result of blade insertion. The data 
therefore can be averaged without extrapolation. 

At 10 feet depth the average b is 4.76 in (0.187 mm ), not 
unreasonable for a silt. At 10.5 feet b is phenomenally high, 12.4 in 
(0.488 mm ), and the data are suspect. However, the data will be 
processed as if nothing were wrong. 



-1 



Depth 



Blade/Orientation 



Cell 



1/E-W 



2/NW-SE 



3/NE-SW 



Aver. 



5-5.5 



10 



10.5 



All 28.0+5.9 21.9+2.5 

24.7 + 1.4 
(omitting 1-2 data) 



21.8 + 1.9 



1 


7.4 


6.1 


2 


6.4 


5.5 


3 


7.0 


6.6 


1 


3.4 


2.7 


2 


3.3 


2.6 



6.6 
7.1 
6.6 

2.5 
2.7 



6.6 + 0.6 



2.9 + 0.4 



The E-W readings are consistently higher, so these data are treated 
separately. Summary values for O, and K are as follows: 



189 



E^W N-S 

Depth v u v h h ° h h_ ^_ 

5-5.5 4.6 4.6 24.7 24.7 5J^_ 21.8 21.8 kTI_ 

10 8.9 1.2 7.7 6.9 5.7 CLJ^ 6.4 5.2 CL7^ 

10.5 8.9 1.2 7.7 3.4 2.2 (LJ3 2.6 1.4 0.2 

w.t. at 7.5'. Assumed y = 125 pcf. 

The k values at 5-5.5 feet are high, indicative of preconsolidation 
in spite of the low stiffness. Possible reasons for the high values are 
as follows: 

(a) The location is a field used by county road maintenance for 
aggregate storage, in which case the soil very well may be preconsolidated. 
A k of 5 suggests the equivalent of about 20 feet thickness of material. 
More likely, the preconsolidation is from heavy equipment. 

(b) The test was conducted in an overconsolidated B horizon. The 
low stiffness suggests this is unlikely. 

(c) The high total stress values are caused by sudden loss of the 
loess structure and excessive pore water pressures. Successive readings 
(averages are reported in the tables) did show a slight tendency to 
decrease with time. A pore pressure transducer integral with the device 
would help answer this question. 

(d) The high stresses are caused by inaccurate determination of 
stiffness parameter b_ , perhaps due to the smothering effect of pore 
water pressure. 






190 



The K values at 10 feet depth are realistic for a normally- 
consolidated silty clay, indicating that preconsolidation of the 
upper layer must have been from equipment with limited contact area 
rather than stockpiled aggregate. 

As previously mentioned, the K values at 10.5 feet are suspect 
because of the abnormally high values of b. The cause is not known, 
but may relate to pore water pressure. 

K values are slightly higher in an E-W direction, as could easily 
occur if preconsolidation was from heavy equipment. 

6. Anita grain elevator data . The first trials with the new 
blade were at the Anita grain elevators site, and included some "de- 
bugging" and loss of data, especially in Boring 1. Boring 2, adjacent 
to a loaded grain bin, and Boring 3, at some distance away, were 
selected for this preliminary analysis. Only the tests at 20 feet depth 
in each hole are analyzed, selected to be within the zone of influence 
of the 42-foot diameter loaded bin, but below any major preconsolidation 
effect from heavy equipment. 



191 



Averaged data are as follows: 



Blade 



Boring Depth Cell 



20' 1 24.5 23* 25 

2 (27) 26 22.5 

3 25 28 26 

20.5' 1 37.5 25 35 

2 35 36.5 28 



24. 


.5 


(27] 


i 


25 




37. 


5 


35 




21. 


5 


(23. 


5) 


28 




24. 


5 


23. 


5 



20' 1 21.5 17.5 20 

2 (23.5) 20 20.5 

3 28 21.5 25 

20.5' 1 24.5 18 19.5 

2 23.5 22 25 



* 



Blade 2 in Boring 2 faced the grain bin. 



2 
Regressions give the following b and r values 



Blade 

Boring 



epth 


1 


2 


3 


20' 


0.32(0.04) 


3.15(0.98) 


0.63(0.07) 


20.5' 


- ( - ) 


12.1K - ) 


- ( - ) 


20' 


4.23(0.97) 


3.29(0.97) 


3.57(0.83) 




- ( - ) 


6.42( - ) 


7.95 ( - ) 



Averaging for four underlined values gives b = 3.56 + 0.48 in. 
(0.14 + 0.02 mm ), which appears to be reasonable. Application of this 



192 



b value gives the following pressures : 



Blade 



Boring Depth Cell 



20' 15.7 14.7 16.0 

12.9 



20.5' 24.0 16.0 22.4 

16.1 



1 




15.7 




14.7 


2 




15.4 




14.9 


3 


Av. 


14.4 


Av. 


14.4 




15.2 


14.7* 


1 




24.0 




16.0 


2 


Av. 


20.1 


Av. 


20.9 




22.0 


18.5* 


1 




13.8 




11.2 


2 




13.5 




11.5 


3 




14.4 




11.0 


1 




15.7 




11.5 


2 


Av. 


13.5 


Av. 


12.6 




14.2 


11.6 



13.3 



Av. 14.1 14.6+1.0 



Av. 19.2 19.9 + 3.3 



20' 13.8 11.2 12.8 

11.7 
12.8 

12.5 
14.3 



Av. 12.8 12.8 + 1.4 



* 



Blade 2 was oriented facing the closest grain bin. + entries 
indicate standard deviations. 



The data show that the highest stress in Boring 2 is not on Blade 
2, facing the close grain bin, but on Blade 1, which is at an angle 
but facing another bin farther away. The difference is not appreciable, 
and less than the anisotropy of the Boring 3 data, presumed to be away 
from the influence of the bins. The full bins exert a nominal floor 



193 



load of 13.5 psi and had a footing design pressure of 13.8 psi. 
Settlements in excess of 6 inches were predicted and appear to have 
occurred. Thus any directional stress anisotropy ordinarily predict- 
able from theory of elasticity may be lost due to plastic deformations 
of the relatively soft alluvial silt. 

Comparison of the Boring 2 and Boring 3 data show a significant 
difference in average horizontal stress, and some stress variations 
with depth in Boring 2. This also is shown in the calculated values 

for k : 
o 

t, t. CT * O . O, O, , K 

Boring Depth v u v h h o 



2 


20' 


17.6 


0.8 


16.8 


14.6 


13.8 


0.82 




20.5' 


17.6 


0.8 


16.8 


19.9 


19.1 


1.14 


3 


20-20.5' 


17.6 


3.5 


14.1 


12.8 


9.3 


0.66 



Water table at 18.5' in Boring 2; 12.1' in Boring 3. Assumed 
Y = 125 pcf. 

We may conclude that the silt adjacent to the grain bin is over- 
consolidated but shows no significant directionality in horizontal 
stress. The value of K in Boring 3 away from the bin is character- 
istic of a normally consolidated fine-grained alluvium. At 20' depth 
10 feet from a 20' radius surface load, the elastic solutions of Foster 

and Ahlvin indicate influence coefficients I ~ 0.13 and I, ~ 0.13, 

u h 

which suggests the loaded bin increases both horizontal and vertical 
stresses by about 1.8 psi. Since this was not taken into account for 



194 



calculation of the vertical stress, it may be subtracted from a ' 

n 

(assuming u from the standing water level is a valid measurement) to 
give pre-load Boring 2 stresses: 

Depth 20', ' = 13.8 - 1.8 = 12.0 psi 

20.5 1 , a ' = 19.1 - 1.8 = 17.3 psi 
n 

The first value is very close to the 12.8 psi measured away from 
the bin. The second is high, perhaps reflecting an uneven layer 
transmission of horizontal stress, i.e. a stiff er layer. This also is 
indicated by the b values in a preceding table. 

7 . Conclusions of Trials with the 3-Blade Stepped Vane . 

(a) Properly functioning pneumatic pressure cells calibrate with a 
factor of 1.0 against air pressure. 

(b) The thinner blade thicknesses in this vane successfully avoid 
problems of bad readings from the thicker (1/4 in.) blades. Yet the 
thin blades have sufficient structural strength because of integration 
into a 3-bladed vane. (The thicker single blade experienced some 
bending with resulting error.) 

(c) Horizontal stress data and stiffness values b with the new 
vane were reasonable from tests of stiff, overconsolidated clay, soft 
loess soil, and alluvial silts. With the soft loess soil the stiffness 
parameter b appears to be zero, and stresses may be appreciably in- 
fluenced by excess pore water pressure developed from pushing the blade. 



195 



(d) A theoretically correct stress increase was measured next 
to a recently loaded structure, but no appreciable directional stress 
anisotropy. This is believed due to non-elastic behavior from large 
settlement of the structures. 

8. Recommendations for future research . 

(a) More field trials, preferably with comparative data, are 
needed . 

(b) Pore pressure transducer (s) should be added to the device. 

(c) Because of the high costs of fabrication, consideration should 
be given to an all- electrical measurement system using diaphragm strain 
gages. 

(d) b values should be related to soil data such as 
penetration test data, etc. It appears likely that empirical relation- 
ships exist, based on preliminary observations that higher b means 
stiff er soils. 

(e) Finally, it should be emphasized that since until now the 
available methods for in situ stress determination have been elaborate, 
time-consuming, and almost prohibitively expensive, virtually every new 
test is a research report, because it gives new insight into soil 
behavior relevant to soil mechanics solutions, soil genesis, and under- 
standing of soils. 



196 



Appendix D 
Disassembly of the 3-Bladed Vane 

1. Extreme care should be used not to allow foreign material to 
get into the gas lines in the blade. If a line (usually the return 

line) does become plugged, try blowing it out with gas pressure. If 
this does not clear the line, repair must be done in the machine shop: 

a. Pry out the inner porous disc from the plugged cell. This 
ruins the disc, which must be replaced. Sometimes the plugging material 
collects in the manifold part under this disc. 

b. Try introducing a thin, flexible wire through the plugged line. 

c. If these measures fail, the blade will have to be re-opened on 
a milling machine for repair. This is about a 1-2 day job for a skilled 
machinist. Lines in the blade are stainless steel hypodermic needle 
stock. 

2. To replace a blade or remove it for repair, the following 
procedure is used: 

a. Unscrew and remove the tip. 

b. Remove 6 screws holding the three hardened steel lead blades, 
and remove the lead blades. 

c. Remove the axial center screw from the upper end. The head of 
this screw is inside of the female recepticle for the drill rod. 

d. Remove six downward-oriented screws holding the blade cap. 



197 



e. Remove six laterally-oriented screws holding the three 
rectangular blade head pieces just under the cap. 

f. Remove all screws from the gusset strips between the blades. 

g. Pull the blades free from the central spine. 

3. Reassembly requires some extra steps in order that all components 
be axially tight against each other: 

a. Install the three hardened steel lead blades on the spine. Leave 
the screws slightly loose. 

b. Line up the three main blades on the spine, install gusset strips 
and screws, and leave the screws slightly loose. 

c. Screw on the tip and tighten. 

d. Tighten all screws previously left loose. 

e. Install blade head pieces, cap, and center screws. 




a 


m 


ifl CD 


— "> 


r+ c+ 


— o 


-5 -T> 


30 \) 


CD "I 


• 


;n 3 


>> 


i.n i- 1 - 


LO 


D 




3 f+ 

i i. 


D 




FEDERALLY COORDINATED PROGRAM (FCP) OF HIGHWAY 
RESEARCH AND DEVELOPMENT 



The Offices of Research and Development (R&D) of 
the Federal Highway Administration (FHWA) are 
responsible for a broad program of staff and contract 
research and development and a Federal-aid 
program, conducted by or through the State highway 
transportation agencies, that includes the Highway 
Planning and Research (HP&R) program and the 
National Cooperative Highway Research Program 
(NCHRP) managed by the Transportation Research 
Board. The FCP is a carefully selected group of proj- 
ects that uses research and development resources to 
obtain timely solutions to urgent national highway 
engineering problems.* 

The diagonal double stripe on the cover of this report 
represents a highway and is color-coded to identify 
the FCP category that the report falls under. A red 
stripe is used for category 1, dark blue for category 2, 
light blue for category 3, brown for category 4, gray 
for category 5, green for categories 6 and 7, and an 
orange stripe identifies category 0. 

FCP Category Descriptions 

1. Improved Highway Design and Operation 
for Safety 

Safety R&D addresses problems associated with 
the responsibilities of the FHWA under the 
Highway Safety Act and includes investigation of 
appropriate design standards, roadside hardware, 
signing, and physical and scientific data for the 
formulation of improved safety regulations. 

2. Reduction of Traffic Congestion, and 
Improved Operational Efficiency 

Traffic R&D is concerned with increasing the 
operational efficiency of existing highways by 
advancing technology, by improving designs for 
existing as well as new facilities, and by balancing 
the demand-capacity relationship through traffic 
management techniques such as bus and carpool 
preferential treatment, motorist information, and 
rerouting of traffic. 

3. Environmental Considerations in Highway 
Design, Location, Construction, and Opera- 
tion 

Environmental R&D is directed toward identify- 
ing and evaluating highway elements that affect 



* The complete seven-volume official statement of the FCP is available from 
the National Technical Information Service, Springfield, Va. 22161. Single 
copies of the introductory volume are available without charge from Program 
Analysis (HRD-3), Offices of Research and Development, Federal Highway 
Administration, Washington, D.C. 20590. 



the quality of the human environment. The goals 
are reduction of adverse highway and traffic 
impacts, and protection and enhancement of the 
environment. 

4. Improved Materials Utilization and 
Durability 

Materials R&D is concerned with expanding the 
knowledge and technology of materials properties, 
using available natural materials, improving struc- 
tural foundation materials, recycling highway 
materials, converting industrial wastes into useful 
highway products, developing extender or 
substitute materials for those in short supply, and 
developing more rapid and reliable testing 
procedures. The goals are lower highway con- 
struction costs and extended maintenance-free 
operation. 

5. Improved Design to Reduce Costs, Extend 
Life Expectancy, and Insure Structural 
Safety 

Structural R&D is concerned with furthering the 
latest technological advances in structural and 
hydraulic designs, fabrication processes, and 
construction techniques to provide safe, efficient 
highways at reasonable costs. 

6. Improved Technology for Highway 
Construction 

This category is concerned with the research, 
development, and implementation of highway 
construction technology to increase productivity, 
reduce energy consumption, conserve dwindling 
resources, and reduce costs while improving the 
quality and methods of construction. 

7. Improved Technology for Highway 
Maintenance 

This category addresses problems in preserving 
the Nation's highways and includes activities in 
physical maintenance, traffic services, manage- 
ment, and equipment. The goal is to maximize 
operational efficiency and safety to the traveling 
public while conserving resources. 

0. Other New Studies 

This category, not included in the seven-volume 
official statement of the FCP, is concerned with 
HP&R and NCHRP studies not specifically related 
to FCP projects. These studies involve R&D 
support of other FHWA program office research. 



DOT LIBRARY 



D0057175